Kinetics of aluminum-fulvic acid complexation in acidic waters

Jun 1, 1987 - Brian J. Plankey, Howard H. Patterson. Environ. Sci. Technol. , 1987 ... F. C. Wu, R. B. Mills, R. D. Evans, and P. J. Dillon. Analytica...
2 downloads 0 Views 981KB Size
Environ. Sci. Technol. IQ87, 21, 595-601

Kinetics of Aluminum-Fulvic Acid Complexation in Acidic Waterst Brian J. Plankey and Howard H. Patterson* Department of Chemistry, University of Maine, Orono, Maine 04469 ~~

~~

A fluorescence technique has been used to study the complex formation kinetics of aluminum with a single metal-free fulvic acid isolated from an Adirondack Mountain forest soil. In the pH range of 3.0-4.5, two kinetically distinguishable components of the fulvic acid mixture have been identified, which define two types of average aluminum binding sites. Both fulvic acid average sites conform to a bidentate chelating binding site kinetic analysis, from which rate and equilibrium parameters have been obtained. From comparison'of rate and equilibrium constants of aluminum-salicyclic acid complexation, we conclude that the faster reacting component of fulvic acid probably contains salicylic acid type aluminum binding sites. Results are also compared with those of an aluminum-fluoride kinetic study. Fulvic acid and fluoride react with aluminum by the same mechanism and therefore have the same pH dependence. The dependence of the rate on temperature is, however, quite different for the two reactions. The environmental implications of these findings are discussed.

Introduction An important consequence of acid precipitation in northeastern North America is increased aluminum mobilization resulting in elevated levels of dissolved aluminum in surface waters (1-3). Increased aluminum concentrations in acidified surface waters have been correlated with increased toxicity to a variety of aquatic plants and animals (1,4). However, aluminum toxicity to some plants and animals is very much dependent on the speciation of the aluminum present in solution ( I , 4 ) . For example, dissolved inorganic complexes of aluminum, such as hydroxides and fluorides, have been shown to be toxic to young fish, whereas complexes of aluminum with naturally occurring organic ligands, such as humic substances, appear to be essentially nontoxic (1,5). Humic substances are the major organic constituents of soils (6) and are among the most important complexing ligands of metal ions in natural waters (7,8). On the basis of solubility, humic substances are generally divided into three fractions: humin, humic acids, and fulvic acids. Fulvic acids are the acid- and base-soluble natural soil organic acid mixtures that have frequently served as laboratory models for organic ligand complexation in natural waters. Over the last 20 years, the metal complexation of fulvic acids has been extensively studied. Nearly all of these studies have focused on the capacity of fulvic acids to bind metals and the thermodynamic stability of metal-fulvic acid complexes. Kinetic studies of metal-fulvic acid complex formation are scarce. Yet, if the influence of natural soil organic matter on the transport, speciation, and toxicity of such metals as aluminum is to be understood, the rate of metal-organic complex formation must be determined. In this investigation, the kinetic and the equilibrium behavior of the complex formation reaction of aluminum with a single metal-free soil fulvic acid (FA) were studied in the pH range of about 3.0-4.5. This is roughly the pH This is a contribution from the ALBIOS project, EPRI Contract RP-2365-01. 0013-936X/87/0921-0595$01.50/0

Table 1. Analytical Characteristics of Adirondack Mountain 0, Horizon Soil Fulvic Acid (FA)

characteristics

FA

characteristics

FA

carbon 49.52 carboxyl COOH,mequiv/g 7.9 hydrogen 4.40 phenolic OH, mequiv/g 3.5 nitrogen 1.47 E,/EG 13.2 sulfur 0.39 % ash 2.6 total acidity, mequiv/g 11.4 g/mol 740 % % % %

z,

range of soil waters and first-order drainage waters in forested watersheds of northeastern North America that are impacted by acidic deposition. The objectives of the investigation were as follows: (i) to quantify the rate of aluminum-FA complexation as a function of pH, temperature, and concentration; (ii) to derive an overall rate expression for the complexation reaction; (iii) to identify a mechanism consistent with the experimental data; and (iv) to determine whether there are ecologically significant rate limitations for the reaction that might affect the biological impact of toxic aqueous aluminum during the mixing of natural waters.

Experimental Section Materials. The FA sample was isolated and purified from the 0, horizon of a forest spodosol in the Huntington Watershed of the Adirondack Mountains of Ne-7 York State by a modification of the method of T h u r r m and Malcolm (9). A total of 1000 g of soil was slurried with 2 L of 4 M NaOH. Nitrogen was bubbled through the slurry, which was covered and allowed to sit for 48 h. The entire solution was centrifuged 10 min at 5000 rpm to remove particulates. The syrupy, black decantate was diluted 101 with distilled water, and the pH was adjusted to 1 with concentrated HC1 to precipitate humic acid. Concentration, purification, and isolation of the soluble FA were accomplished by adsorption chromatography, cation-exchange chromatography, and lyophilization, respectively, as described by Thurman and Malcolm (9). The cation-exchange procedure was used to produce a metal-free sample in the fully protonated form. Carbon, hydrogen, nitrogen, and sulfur analyses were performed by Galbraith Laboratories, Knoxville, TN. Found C, 49.52; H, 4.40; N, 1.47; S, 0.39 (on a dry, ash-free basis). Moisture content was found to be 2.91% by drying to constant weight under vacuum at 60 "C. The ash content was determined to be 2.6% by heating a 0.050-g sample in a crucible to constant weight with a Fisher burner. Total acidity and carboxyl functional group analyses were determined by the methods of Schnitzer and Gupta (10). The E,/& ratio was measured at pH 8 as described in the literature (11). A number-average molecular weight determination of our FA sample gave 740 by vapor-pressure osmometry in tetrahydrofuran (THF) at 45 "C. The four-point determination, extrapolated to zero concentration, was performed by Huffman Laboratories, Golden, CO. Results of the analytical determinations are in Table I. Aluminum solutions were prepared fresh each day from reagent-grade A1(N03)3-9H20.Initial aluminum concenM. Initial trations varied between 9.0 X lo4 and 3.6 X

0 1987 American Chemical Society

Environ. Sci. Technol., Vol. 21, No. 6, 1987 595

FA concentrations varied between 5 and 10 mg/L. With 740 for the molecular weight of FA, aluminum was always in a t least a 130-fold excess over FA. Although the FA molecular weight is subject to some error (12), the aluminum concentration is almost certainly always large enough for pseudo-first-order kinetics to be in effect (13). Whereas for many purposes it may be better to report results on a milligrams per liter basis for fulvic acid, molar units are used here because of the convenience and clarity in reporting kinetic results. Hence, with a M nof 740, initial FA concentrations varied from 6.8 X lo4 to 1.4 X M. All fulvic acid solutions were prepared from 1.4 X lo4 M (100 mg/L) standard solutions, which were filtered through a Nucleopore filtration apparatus with 0.4-wm polycarbonate filters prior to reaction to remove any particulate matter. All solutions were buffered with either sodium acetate-acetic acid mixtures or sodium monochloroacetatemonochloroacetic acid mixtures, and the ionic strength was adjusted to 0.1 M with sodium chloride. It has been shown that neither acetate nor monochloroacetate nor chloride forms complexes with aluminum, to any significant extent, in the range of pH and concentration used in this study (14-16). Several different pHs were studied ranging from about 3.0 to 4.5. Solution pH was determined by a Fisher universal glass electrode with a Fisher saturated calomel reference electrode connected to an Orion digital pH meter. Standard buffers were used for reference with no corrections made for liquid junction potential differences. The concentration of H+ was taken as [H+]= 10-PH/yH+,with yH+= 0.78, calculated from the Davies equation (14),where the square brackets denote concentration in moles per liter. The concentrations of all species were determined as molarities rather than activities. However, since the ionic strength was constant throughout, the molarity of each species differed from its activity by only a proportionality constant dependent on the charge of the species. Kinetic Procedure. A fluorescencetechnique was used to follow the course of the reactions on the basis of the finding that fulvic acid fluorescence intensities increase upon binding to aluminum. The wavelengths of excitation (350 nm) and emission (400 nm) a t which the difference between the fluorescence intensity of free fulvic acid and the fluorescence intensity of the aluminum-FA complex is at a maximum were determined by synchronous scan fluorescencespectroscopy (I7). Wavelengths for excitation and emission were tested from 270 to 610 nm. With aluminum in large excess, fluorescence intensities increased on the average of about 85% in our experiments. We also found an increase of roughly 1000% in the fluorescence of salicylic acid upon reaction with a large excess of aluminum. In fact, using our fluorescence technique we were able to satisfactorily reproduce the kinetic data for the reaction of salicylic acid with aluminum studied previously, where absorbance was used to monitor the course of the reaction (16).For this reason, plus the fact that other researchers have found agreement between fluorescence and ion-selective electrode methods in metal binding studies with FA in which no coagulation or precipitation was present (18-20), it was assumed that increased fluorescence intensity was proportional to aluminum-FA complexation. During the course of the reactions reported here and in solutions that had reacted for several weeks, there was never any sign of coagulation or precipitation either by visual examination or by light scattering a t 350 nm. Kinetic runs were carried out with a stopped-flow technique. Buffered and thermostated solutions of FA and 596

Environ. Sci. Technol., Vol. 21, No. 6 , 1987

-1 -7

-8

0

10

20

30

40

50

60

TIME, s

Figure 1. Pseudo-first-order kinetics plot for reaction of aluminum with M and [FA] = FA at pH 3.32 and 25 'C: [Al(III)]total= 1.8 X 6.8 X M.

aluminum were drawn into two separate syringes, and the plungers were pushed together so as to deliver equal amounts of each solution to a mixing chamber and then directly into a thermostated quartz spectrofluorometer flow cell for observation. The course of the reactions was followed by monitoring the change in intensity of emitted light at 400 nm upon excitation at 350 nm. Fluorescence intensities were determined with a Perkin-Elmer Model MPF-44A spectrofluorometer connected to a strip chart recorder and stored digitally on a Bascam-Turner Model 4120 X-Y recorder disk. For each run, 500 fluorescence intensity values were recorded and stored a t equal time intervals, which were varied according to the speed of the reaction. Treatment of Data. Infinite time fluorescence intensities were determined for each run, and pseudo-first-order plots of In ( I , - I t ) vs. time were obtained directly from the Bascam-Turner processor-plotter. In an attempt to obviate infinity readings, Guggenheim plots of In (It+AtI t ) vs. time were similarly made where no experimental infinity values are necessary (21,22). While these plots were qualitatively the same as the In (I, - It) vs. time plots, they proved to be too noisy to extract useful quantitative data. Therefore, actual measured infinite time intensities were used in In ( I , - I t ) vs. time plots. Both the In (It+At - It) vs. time Guggenheim plots and the In (I, - I,) vs. time kinetic plots were used to objectively determine the proper number of straight-line components. It was believed that two straight-line portions best described all of the pseudo-first-order kinetic plots. These plots are then consistent with two first-order reactions in which the observed rate parameters are separated by nearly an order of magnitude or more (22). The kinetic data can then be treated by the expression I = A[1 - &b*dlt] +- B[1 - e-kobsd2t] + x (1) where I is the fluorescence intensity, hobsdl and hob& are the observed pseudo-first-orderrate parameters for the two reactions, X is a time-independent term including blank and fast reacting components, and A and B are constants (12). Equation 1 can be treated satisfactorily by a nonlinear regression, provided reliable estimates of the values of the parameters to be fitted can be obtained (22). We obtained estimates for hob& and B from the slopes and intercepts, respectively,of the long-time linear portion of the In (I -m I t ) vs. time plots (see Figure 1) (23). EStimates for kobsdIand A were obtained from the slopes and

Table 11. Observed Rate Parameters for Reaction of Al(II1) with Fulvic Acid

[H'], M 1.42 x 1.42 X 1.42 X 1.42 X 7.82 x 7.82 X 7.82 X 6.07 x 6.09 X 6.67 X 6.67 X 2.02 x

10-3 loT3

10-4 10-4

10-4 2.21 x 10-4 2.42 X 2.42 X 1.42 x 1.42 X 6.07 x 6.09 X 6.67 X 2.02 x

10-3 loT3

10-4 loy4

10-4

[Al(III)]" X lo3, M

[FA] X IO6, M

1.8 2.7 3.6 3.6 1.8 2.7 3.6 1.8 2.7 3.6 3.6 1.8 2.7 3.6 3.6

25 " C 6.8 6.8 6.8 14 6.8 6.8 6.8 6.8 6.8 6.8 14 6.8 6.8 6.8 14

1.8

2.7 1.8 2.7 3.6 1.8

15 " C 6.8 6.8 6.8 6.8 6.8 6.8

kobsdlr

4.85 X 5.48 X 6.08 X 6.01 X 6.25 X 7.69 X 8.73 X 8.84 X 1.09 X 1.31 X 1.29 X 1.49 X 1.86 X 2.11 X

2.08 X 4.14 X 4.53 X 8.16 X 9.79 X 1.14 X 1.29 X

s-'

kobsdz, 8-l

6.24 X 6.71 X 7.23 X 7.26 X lo-' 8.94 X 1.24 X 1.35 X 1.32 X 10-1 1.69 X lo-' 2.02 X lo-' 2.06 X lo-' 1.94 X lo-' 2.42 X lo-' 3.20 X 10-1 3.00 X lo-' lo-' lo-'

lo-'

IO-' lo-' lo-' lo-'

lo-' lo-' lo-' lo-' lo-'

4.73 X 5.12 X 1.00 X lo-' 1.30 X lo-' 1.46 X lo-' 1.57 X

IAl(1II)l = total aluminum concentration.

intercepts, respectively, of In A vs. time plots according to eq 2 (23). For each run, 500 data points were transA I - I , - Be-komzt = Ae-kobsdlt (2) ferred directly from disk to a mainframe computer where a standard nonlinear least-squares computer program based on the Gauss algorithm SAS-NLIN was run. The program fits the intensity-time data and refines the estimated parameters hobs&, kobd2, A , and B according to eq 1.

Results Observed Rate Parameters. The reactions of aluminum with FA followed pseudo-first-order kinetics because [A13+]was large compared to [FA] and the solutions were buffered. All of the pseudo-first-order kinetic plots consisted of two linear segments with different slopes, as can be seen in Figure 1. This is in accord with several previous studies of metal-FA interactions that have postulated two general types of metal binding sites on FA (24-26). In addition, kinetic studies of the dissociation of iron(II1) and aluminum(II1) complexes of FA have indicated the existence of two metal-FA components (12,27). The nonlinear least-squares treatment of estimates from the In ( I , - It) vs. time plots yielded two observed rate parameters, as described under Experimental Section. These koWl and hob& values are shown in Table 11. Each is the average of at least five runs. The mean standard deviation for the hobdlvalues was 7.6%, while for the kobd2 values it was 9.3%. With [H+]about 1.9 X M and less (pH 3.8 and above), the reaction was too fast to obtain reproducible hobad values with the stopped-flow fluorescence technique. Equation 1, from which kobsdl and kobsd2 were obtained, is exact for well-defined mixtures but in the case of a complex mixture such as FA becomes an approximation (27). This is so because FA has a distribution of different binding sites (12, 27) consisting of oxygen-containing functional groups, particularly COOH and OH (28). While no two binding sites are necessarily chemically identical (29,24),the two straight-line segments of the kinetic plots

indicate that there are two kinetically distinguishable components of the FA mixture. Therefore, hobdl and koM2 are in reality averages over two kinetically distinguishable parts of the distribution of aluminum binding sites of FA (12,27),and A and B in eq 1 are the fluorescence intensities proportional to the initial concentrations of the two kinetically distinguishable components of FA. A further consequence of the complex mixture nature of FA is that an increased strength of metal binding occurs with decreasing metal to ligand ratios (30). Therefore, metal-FA rate and equilibrium parameters cannot in any way be viewed as constants; they depend, among other things, on the degree of metal coverage. For this reason many researchers have recently sought to model metal-FA interactions by considering FA as sufficiently complex to be represented by a continuous distribution of metal binding sites (24,30,31). A major theme of these studies has been to reject the notion that FA complexation can be understood in terms of simple monomeric ligand models. Instead, these studies have stressed the idea that rate and equilibrium constants can exist only as differential quantities. Even more recently, however, this continuum model has been criticized as an overestimation of the complexity of the problem (32). It is our view that the truth probably lies somewhere between the two extremes of discrete and continuous binding sites. Since chelating sites similar to salicylic acid (2hydroxybenzoic acid) are commonly thought to be present in FA (12, 24, 29) and aluminum-salicylic acid kinetics have been extensively studied (14, 16, 33), salicylic acid appeared to be a reasonable model with which to test our FA kinetic data. It should be emphasized that our intent was not to choose a model a priori but rather to see if our data, which had indicated two general types of kinetically distinguishable FA-aluminum binding sites, would conform to a previously reported aluminum-salicylic acid kinetic analysis (14). If so, our secondary goal was to then compare any resulting aluminum-FA rate and equilibrium parameters to aluminum-salicylic acid parameters for chemical reasonableness, in consideration of possible FA binding sites. Third, we desired to draw conclusions about the chemistry of natural waters, where possible. Equilibrium Parameters. On the basis of the general mechanism for the reaction of Al(II1) with chelating salicylic acid ligands (14, 16, 33), we propose the following reaction scheme for the reaction of Al(II1) with supposed chelating sites on FA: H ~ Ot H+ t AIS+ t HL-

It

k,

.

H~ t A I O H ~ +t H ~ L

JI. . AlOHL t 3 H f

2Ht t AIOH2* t HL-

A deprotonated chelating site on FA has been designated as HL-, and H2L represents a fully protonated site. Coordinated waters have been omitted for simplicity. We have assumed a 1:1 stoichiometry for these reactions for the following reasons. First, aluminum was always in large excess over FA, and therefore complexes in which the FA:A1 ratio was greater than 1:l were highly unlikely. Second, there were no signs of precipitation or coagulation to indicate the formation of chains with aluminum bridges between FA molecules. Third, at p = 0.1, 1:l complexes Environ. Sci. Technol., Vol. 21, No. 6, 1987

597

I

5 1

't 0 0

2

4

6

Flgure 2. Kinetic results of of [H'] at 25 OC.

1 0 1 2 1 4

8

[H']

X IO4, M

[H']I[K,,,([H+]

+ KAlw)]as a function

~

"0

I

2

3

4

[Al(lD)] X IO3, M.

have been determined for FA with several metals including iron, copper, and aluminum (34). In analogy with the reaction of aluminum with a salicylic acid (141, we define a formation parameter for the aluminum-FA complex from AP+ and a deprotonated chelating site on FA: k

AI3+ + HL- & AlL+ + H+ k-i

(3)

The formation parameter is K1 = kl/k-l = [AlL+][H+]/[Al3+][HL-]

Figure 3. Observed rate constant kM, as a function of total aluminum concentration at pH 2.95 and 25 O C .

Because all of the forward and reverse reactions of the proposed reaction scheme are pseudo first order under the conditions used in this study, the observed rate parameter can be,derived in analogy with relaxation kinetic experiments (14). Assuming the protolytic equilibria represented by the vertical arrows in the proposed reaction scheme are rapid in comparison to complex formation (14, 16),kobsd has been derived (14) in the form

(4)

Likewise, an apparent equilibrium parameter Kappis defiied as the formation parameter of the AlL+ complex from total aluminum and total ligand (14, 16): Kapp = [AlL+][H+]/[([AP+]+ [A10H2+])([H2L+ HL-I)] (5) Substituting the hydrolysis constant of aluminum, KMOH, and the first acid dissociation parameter of the supposed chelating group on FA, KHL, gives Kapp = KlKHL[H+I/[(KAlOH + [H+l)(KHL+ ["I)]

(6)

Rearrangement yields

hKHL

+ kZKAlOH + k3[H+I +

with B = [AI(III)] + [H+]/Kapp

(11)

From eq 8, 9, and 11 we see that (12) Rearrangement of eq 10 and substitution of eq 12 yields = kobsd/kf

..

Equation 7 was tested for the faster reacting kinetically distinguishable component of our FA by determining Kapp values from hobsdl values (see below) and plotting the left-hand side against [H+],as shown in Figure 2. The hydrolysis constant of aluminum at ionic strength 0.10 M, KAloH = 3.55 X lo4 M, was obtained from the literature (14). From the slope and intercept of Figure 2 were obtained KHL = 3.57 X M and K, = 0.346 for the faster reacting component of our FA. Kinetics. For both kobsdl and kobsd2,since all of the forward and reverse reactions are pseudo first order, at constant [H+], kobsd is a linear function of the total aluminum concentration [Al(III)]: kobsd

= kdA1(111)1

kr[H+l

(8)

kf and k, are the forward and reverse rate parameters for the overall reactions (14). Furthermore (14) Kapp = %/kr (9) The koMl values along with eq 8 and 9 were used to obtain the K, values used above in eq 7. A typical kobsdl vs. [Al(III!] plot at constant [H+] is shown in Figure 3. 598

Environ. Sci. Technol., Vol. 21, No. 6, 1987

(13) Figure 4 is a plot of the left side of eq 13 vs. l/[H+] calculated with the kobsdl values of the faster reacting component of our FA. A fairly good straight line was obtained, indicating that the k3[H+]term was very small or zero. The same was found for the reactions of aluminum with salicylic acids as well (14, 33). From the slope and intercept of Figure 4 were obtained the values for the faster reacting component of k4KHfiMOH = 4.43 X loi M s-l and S-'. klKHL + k & ~ l o=~ 1.21 x The same aluminum-salicylic acid kinetic analysis was performed with the kobsdl values of the faster reacting component of our FA at 15 " C for a smaller number of observations. Table I11 shows the equilibrium and kinetic parameters obtained at both 25 and 15 "C. The k4 values were calculated from the values of k 4 K ~ ~ Kwith ~ 1 0KHL ~ derived at each temperature and with the literature value for KAlOw Because the reactions with forward rate parameters kl and k2 have the same pH dependence, the sum of klKm k2KA10H cannot yield separate values of kl and k2 without making an assumption about one term or the other (33) (see discussion below). From the temperature dependence of the k4 values were obtained the activation

+

I

0

0

6

12

18

I/["]

30

24

X IO-',

36

42

48

M-'

Figure 4. k,KIKHLIKapp as a function of at 25 O C .

1/[H+]from kobsdl Values

Table 111. Equilibrium and Rate Parameters Derived from k o b @ d lValues of Al(II1)-FA Reaction at 25 and 15 O C

parameter KHL,M K1 klKHL + kZKAiOH, S-l ~ ~ K H J I A IM o H8-l, ka, M-' s-l

25 " C 3.57 x 0.346 1.21 X 4.43 x 3.50 x

10-3 lo-' 10-5 103

15 " C 3.23 x 10-3 0.264 1.89 X 3.09 x 10-6 2.69 x 103

6

12

18 24 I / [H'] X

30

36

42

48

M-'

Figure 5. k,KIKH,/K,ppas a function of I/[H+] from kObsdP values at 25 O C .

Table V. Comparison of Rate Parameters of Fulvic Acid Components at 25 "C

rate parameter

from kobsdl

from kobsdZ

k,, M-' s-l

2.30 1.10 x 103 3.50 x 103 6.65

0.458 1.25 X lo2 4.24 X lo2 1.73

k,, M-l 5-l kq, M-l 5-l k-1,

&I-'

5-l

upper limit for k l of 3.39 M-l s-l. In the same way, we obtained an upper limit for kz of 3.41 X lo3M-l s-l. These values can be compared with upper limits of kl and k 2 for salicylic acid of 2.5 M-l s-l and 1.4 X lo3 M-l s-l , r especTable 1V. Equilibrium and Rate Parameters Derived from tively (33). k o b t d l Values of Al(JI1)-FA Reaction at 25 OC Further refinements can be made in assessing the relative contributions of the reactions with forward rate paparameter value parameter value rameters kl and kz by envoking the Eigen mechanism (35) KHL,M 2.59 X k q K ~ & ~ l oM ~ , S-l 3.90 x lo4 as was done in our earlier paper (36). Doing so yielded a Kl 0.240 kq, M-'s-l 4.24 X lo2 value of kl = 2.30 M-' s-l and the conclusion that both krKwr + k,KAl,-,u, 1.63 X reactions contribute to the overall reaction. Another test for the validity of the rate parameter values reported here parameters for reaction 14. They are AH* = 3.91 kcal/mol was to calculate k4 according to the Eigen mechanism (35), and AS* = -29.2 eu. with the ligand penetration rate constant for The proposed reaction scheme may represent the (H20)5A10H2+ obtained in our previous paper (36). This mechanism for the slower reacting kinetically distingave k4 = 3.71 X lo3 M-I s-l, in good agreement with the guishable average binding site, represented by the kobsdz observed value of 3.50 X lo3 M-' s-l. values, as well as for the faster reacting average site. In We next consider the rate parameter values obtained order to test this, we performed the same kinetic treatment from the kobada values for the slower reacting average as above using the k&dZ observed rate parameters in place binding site of our FA. Making the same assumptions as of the howl values. Figure 5 shows the final kfKIKHL/Fapp in the case of the faster reacting average binding site, vs. l/[Ht] plot, which again yielded a fairly good straight approximate values of k l , k,, k4, and k_l were obtained for line. The equilibrium and kinetic parameters derived from the average slower reacting site as well. Table V shows the kobsd2 values for the slower reacting average binding a comparison of the rate parameter values for the two site at 25 "C are shown in Table IV. kinetically distinguishable components of our FA. Equilibrium Parameters. The equilibrium parameter Discussion values obtained for reaction 3 for both kinetically distinRate Parameters. The rate parameter k4 for the faster guishable components are Kl,aitel= 0.346 and Kl,site2= reacting average aluminum binding site of our FA can be 0.240. These can be compared with the values for salicylic compared to literature values of the same reaction with acid, 5-sulfosalicylic acid, and 5-nitrosalicylic acid of 1.33, salicylic and substituted salicylic acids. We obtained k4 7.60, and 13.0, respectively (33). It should be reemphasized = 3.50 X lo3 M-l s-l from the kobsdl values at 25 "C and that because of the complex mixture nature of FA, the p = 0.10 M. Under the same conditions, the values of k4 aluminum-FA kinetic and equilibrium parameters are not obtained for salicylic acid, 5-sulfosalicylic acid, and 5-niconstants. Since we have used a large excess of aluminum trosalicylic acid are 3.90 X lo3, 2.50 X lo3,and 1.54 X lo3 over FA in our studies, the degree of coverage of FA M-l s-l, respectively (33). The two terms of the intercept binding sites will be quite high. Therefore, the average of Figure 4, klKHL+ kaAIOH,cannot be separated because values of forward rate parameters and equilibrium pathe reactions with forward rate parameters k l and k2 have rameters are likely to be greater (and reverse rate paramthe same pH dependence. Assuming k2 is very small or eters smaller) in cases where AkFA ratios are reduced (24). zero and the total intercept is equal to klKHLyields an The high degree of metal coverage in our studies could Environ. Sci. Technol., Vol. 21, No. 6, 1987

599

Table VI. Comparison of Acid Dissociation Parameters of Fulvic Acid with Common Organic Acids acid benzoic phthalic salicylic 5-sulfosalicylic 5-nitrosalicylic 3-hydroxybenzoic 4-hydroxybenzoic catechol (1,2-dihydroxybenzene) fulvic fulvic fulvic

KH1

6.46 x 1.58 x 2.04 x 3.75 x 5.70 x 8.7 X 3.3 X 6.03 X

10-3 10-3

10-3 10-3 10-3

ref

KH2

7.94 x 10-6 5 x 1044 8 x 10-13 7 x 10-11 1.2 X 4.8 X 1X

38 39 33 33 33 38 38 40

4.7 x 10-3 to 3.2 x 10-5 to 29 2.5 x 10-3" 2.6 x i o - ~ b 3.57 x 10-3c this work this work 2.59 x 1 0 - 3 d

"Over the pH range 2.76-3.96, type I carboxyl groups. Over the pH range 3.67-4.05, type I1 carboxyl groups. KHL,sltel. dKHL,slte2.

account for the somewhat smaller values of K1 obtained for FA as compared to the values reported for salicylic acids. Two average acid dissociation parameters, Kmaitel= 3.57 x M and KHL,slte2 = 2.59 x M, were obtained from the kobdl and kobd2values, respectively. FA shows two acid titration end points experimentally, commonly called type I and type I1 carboxyl groups (29). Average acid dissociation constants over infinitesmal increments of type I and type I1 carboxyl groups have been defined and calculated for a Canadian soil FA (37). These and values are not constants but are dependent on the degree of ionization, and hence pH (37), as are KHL,sitel and KHblte2. Over the pH range used in this study, the values of K1 and for the Canadian soil FA range approximately from 4.7 X to 2.5 X and from 3.2 X 10-5 to 2.6 X respectively. It appears that both of the kinetically distinguishable sites of our FA that initially (before chelation) bind aluminum are type I carboxyl groups. Table VI shows Km,slteland KHL,sikaalong with literature values of acid dissociation constants for FA and some common organic acids. The average acid dissociation parameter values for both sites indicate that they are most similar to salicylic and phthalic acid sites. Binding Sites. Because the faster reacting kinetically distinguishable component of our FA can be modeled as an average binding site that conforms reasonably well with an aluminum-salicylic acid kinetic treatment and the resulting FA kinetic and equilibrium parameters are in all cases comparable to those found for salicylic acids, we conclude that the faster reacting component of our FA probably contains salicylic acid type aluminum binding sites. The slower reacting kinetically distinguishable component of our FA can also be modeled as an average binding site, which again conforms to an aluminum-salicylic acid kinetic analysis. However in this case, the resulting FA kinetic and equilibrium parameters are smaller than those for salicylic acids. It has been suggested that these slower reacting sites are phthalic acid type sites (12, 24, 41), catechol- (1,2-dihydroxybenzene)type sites (42),or perhaps salicylic acid type sites as well (12, 26). Our data indicate that the slower reacting average binding site contains a type I carboxyl group (29), and the acid dissociation values in Table VI suggest the possibility of either a phthalic acid type site or a salicylic acid type site for this average site. Both an average phthalic acid type site and an average salicylic acid type site would conform to an aluminum-salicylic acid kinetic treatment 600

Environ. Sci. Technol., Vol. 21, No. 6 , 1987

in the pH range of this study. A phthalic acid type site for the slower reacting site would be favored on the basis of Buffle et al. (8) and Gamble et al. (25) finding a ratio of carboxylic to phenolic groups of FA coordinated to copper of about 2:l. However, since there are no aluminum-phthalic acid kinetic or equilibrium data in the literature for comparison with our data, a phthalic acid type site interpretation for the slower reacting average site of our FA must be considered as additionally tentative. Our data, for example, do not rule out the possibility that the slower reacting average binding site is a salicylic acid type site differing from the faster reacting average site by the inductive effects of other substituents on the benzene ring. Environmental Implications. The overall rates under the conditions used by us were such that the first half-life period at 25 "C varied from about 56 s a t pH 2.95 with [A13'] = 1.8 X M and [FA] = 6.8 X M to about 3 s at pH 4.55 with the same initial concentrations. Neglecting for the moment the fact that rate and equilibrium parameters vary with the degree of metal coverage, half-lives were calculated at aluminum concentrations typically found in natural waters. At pH 2.95, 25 "C, and p = 0.10 M, the first half-life period increases to about 230 s with [FA] = 6.8 X M and [A13+]= 2.00 x M. A further extrapolation to p = 0.001 M utilizing the Debye-Huckel theory, at 25 "C and pH 2.95, yields a half-life of about 35 s. We have also determined the temperature dependence of the aluminum-FA reaction. The rate of complexation is fairly insensitive to change in temperature as was indicated by the low activation enthalpy for the reaction with the forward rate parameter k4, AH* = 3.9 kcal/mol. Therefore, a t 0 "C and pH 2.95 we can expect a half-life M and [A13'] = of only about 200 s for [FA] = 5.6 X 2.00 X M at p = 0.001 M. If we now consider that as the ALFA ratio is reduced from about 300-500 used in our studies to 0.36 of the example just cited, the half-life will be reduced as well, because a greater portion of stronger, faster reacting FA binding sites will dominate the averages (24, 29). The rate of aluminum-FA complexation exhibited a strong pH dependence, as was the case for the reaction of aluminum with fluoride ion (36). If the proposed mechanism is valid, eq 10 shows how the observed rate parameters vary with [H']. At high pH k4KHLKAIoH/[H+] dominates, and there was a rapid increase of rate with increase in pH. In fact, above about pH 4.3 there was an even more rapid increase in rate due apparently to more highly hydrolyzed species of Al(II1) providing additional paths (14, 16, 37). Below about pH 2.5, k 4 K ~ L K A I ~ ~ /becomes [H+] small compared to klKHL k2KNOH. And even though k3 is too small to be seen at higher pH, k3[H+]may contribute to an increase in rate as the pH is reduced further still. The net result of the pH dependence of the aluminum-FA reaction is that the overall rate will be a t a minimum around pH 2.5. This was nearly exactly the case found for the reaction of aluminum with fluoride as well (37). We can therefore conclude that, in the absence of effects other than those just discussed, the half-life for aluminum complexation with FA will always be less than 200 s under conditions typically found in natural systems and that equilibrium will be effectively obtained within minutes of the mixing of aluminum- and FA-containing natural waters. An interesting aspect of the aluminum-FA kinetics is the temperature dependence as compared to the aluminum-fluoride reaction. This can be seen most clearly by focusing on just one path in the reaction scheme. For the

+

reaction with forward rate parameter k4 and the analogous reaction of A10H2+with F-, we have determined the forward rate parameter as well as the activation parameters. At 25 "C, the values are It = 3.50 X lo3M-l s-l, AH*= 3.91 kcal/mol, and AS* = -29.2 eu for FA and k = 3.61 X lo3 M-l s-l, AH* = 13.3 kcal/mol, and AS* = 2.29 eu for fluoride (37). As a consequence of these activation parameters, the two forward reactions of A10H2+reacting with F- and FA will be at their isokinetic temperature at 25 "C and will proceed at nearly the same rate (43). As the temperature is lowered, however, the fluoride reaction will be slowed considerably while the fulvic acid reaction will be only slightly decreased in rate (43). This may be particularly significant during periods of acid snowmelt when increased discharge results in increases in both aluminum and hydrogen ion concentrations (1,44) and the aluminum toxicity potential is highest. Fluoride complexes of aluminum have been shown to be more toxic to fish than aluminum-organic complexes ( 1 , 5). Lowered temperatures at this time, however, appear to kinetically favor the complexation of aluminum by FA over complexation by fluoride, thereby resulting in a decrease in the potential for aluminum toxicity. Equilibrium parameter values will also have an effect on whether the complexation of aluminum by either FA or fluoride is favored in any natural system. The equilibrium parameter values reported here for the aluminum-FA complex are approximately 6 orders of magnitude smaller than aluminum fluoride equilibrium constants ( 4 8 , and thus, complexation of aluminum by fluoride is thermodynamically favored over complexation by FA. Again, however, because of the high degree of metal coverage in our experiments, we can expect an increase in equilibrium parameter values for aluminum-FA complexation as the aluminum concentration is reduced to levels found in natural systems (24). In the absence of other effects, the thermodynamic advantage of aluminum-fluoride complexation over aluminum-FA complexation will decrease as the A1:FA ratio decreases.

Acknowledgments We acknowledge the able assistance of Lawrence Allin with our computer programming and Christopher Cronan and Paul Bloom for the review of a preliminary draft of our manuscript.

Literature Cited Driscoll, C. T., Jr.; Baker, J. P.; Bisogni, J. J., Jr.; Schofield, C. L. Nature (London)1980,284, 161-164. Cronan, C. S.; Schofield, C. L. Science (Washington,D.C.) 1979,204,304-306. Johnson, N. M.; Driscoll, C. T.; Eaton, J. S.; Likens, G. E.; McDowell, W. H. Geochim. Cosmochim. Acta 1981, 45, 1421-1437. Burrows, W. D. CRC Crit. Rev. Environ. Control 1977, 7, 167-216. Johnson, N. M.; Likens, G. E.; Feller, M. C.; Driscoll, C. T. Science (Washington,D.C.) 1984, 225, 1424-1425. Schnitzer, M. In Soil Organic Matter; Schnitzer, M.; Kahn, S. U., Eds.; Elsevier Scientific: Amsterdam, 1978; p 1. Langford, C. H.; Gamble, D. S.; Underdown, A. W.; Lee, S. In Aquatic and TerrestrialHumic Materials;Christman, R. F.; Giessing, E. T., Eds.: Ann Arbor Science: Ann Arbor. MI, 1983; Chapter 11. ' Buffle, J.; Greter, F. L.; Haerdi, W. Anal. Chem. 1977,49, 216-222.

(9) Thurman, E. M.; Malcolm, R. L. Environ. Sci. Technol. 1981, 15, 463-466. (10) Schnitzer, M.; Gupta, U. C. Soil Sci. SOC.Am. Proc. 1965, 29, 275-277. (11) Chen, Y.; Senesi, N.; Schnitzer, M. Soil Sci. SOC.Am. J. 1977,41, 352-358. (12) Mak, M. K. S.; Langford, C. H. Can. J . Chem. 1982, 60, 2023-2028. (13) Espensen, J. H. Chemical Kinetics and Reaction Mechanisms; McGraw-Hill: New York, 1981; pp 12-15. (14) Perlmutter-Hayman, B.; Tapuhi, E. Inorg. Chem. 1977,16, 2742-2745. (15) Holmes, L. P.; Cole, D. L.; Eying, E. M. J . Phys. Chem. 1968, 72, 301-304. Secco, F.; Venturini, M. Znorg. Chem. 1975,14,1978-1981. Vo-Dinh, T . Anal. Chem. 1978,50, 396-401. Ryan, D. K.; Weber, J. H. Anal. Chem. 1982,54,986-990. Saar, R. A.; Weber, J. H. Anal. Chem. 1980,52,2095-2100. Cabaniss, S . E.; Shuman, M. S. Anal. Chem. 1986, 58, 398-40 1. (21) Espensen, J. H. Chemical Kinetics and Reaction Mechanisms; McGraw-Hill: New York, 1981; pp 25-26. (22) Mak, M. K. S.; Langford, C. H. Znorg. Chim. Acta 1983, 70, 237-246. (23) Espensen, J. H. Chemical Kinetics and Reaction Mechanisms; McGraw-Hill: New York, 1981; p 67. (24) Gamble, D. S.; Underdown, A. W.; Langford, C. H. Anal. Chem. 1980,52, 1901-1908. (25) Gamble, D. S.; Schnitzer, M.; Hoffman, I. Can. J. Chem. 1970,48, 3197-3204. (26) Bresnahan, W. T.; Grant, C. L.; Weber, J. H. Anal. Chem. 1978, 50, 1675-1679. (27) Langford, C. H.; Mak, M. K. S. Comments Znorg. Chem. 1983, 2, 127-143. (28) Saar, R. A,; Weber, J. H. Environ. Sci. Technol. 1982,16, 510A-517A. (29) Gamble, D. S.; Schnitzer, M. In Trace Metals and Metal-Organic Interactions in Natural Waters;Singer, P. C., Ed.; Ann Arbor Science: Ann Arbor, MI, 1974; Chapter 9. (30) Perdue, E. M.; Lytle, C. R. Environ. Sci. Technol. 1983, 17,654-660. (31) Shuman, M. S.; Collins, G. J.; Fitzgerald, P. J.; Olsen, D. L. In Aquatic and Terrestrial Humic Materials;Christman, F. R.; Gjessing, E. G., Eds.; Ann Arbor Science: Ann Arbor, MI, 1983; Chapter 17. (32) Marinsky, J. A.; Ephraim, J. Enuwon. Sci. Technol. 1986, 20,349-354. (33) Perlmutter-Hayman, B.; Tapuhi, E. Znorg. Chem. 1979,18, 875-877. (34) Schnitzer, M.; Hansen, E. H. Soil Sci. 1970,109, 333-340. (35) Eigen, M.; Wilkins, R. G. Mechanisms of Inorganic Reactions;Gould, R. F., Ed.; Advances in Chemistry 49; American Chemical Society: Washington, DC, 1965; pp 55-80. (36) Plankey, B. J.;Patterson, H. H.; Cronan, C. S. Enuiron. Sci. Technol. 1986,20, 160-165. (37) Gamble, D. S. Can. J . Chem. 1970, 48, 2662-2669. (38) Handbook of Chemistry and Physics, 61st ed.; CRC: Boca Raton, FL, 1980-1981; pp D164-Dl66. (39) Stevenson, F. J. Soil Sci. 1977, 123, 10-17. (40) Motekaitis, R. J.; Martell, A. E. Inorg. Chem. 1984, 23, 18-23. (41) Murry, K.; Linder, P. W. J . Soil Sci. 1984, 35, 217-222. (42) Ephraim, J.; Alegret, S.; Mathuthu, A.; Bicking, M.; Malcolm, R. L.; Marinsky, J. A. Enuiron. Sci. Technol. 1986, 20, 354-366. (43) Thusius, D. Znorg. Chem. 1971, 10, 1106-1108. (44) Johnson, N. M.; Likens, G.; Bormann, F. H.; Fisher, D. W.; Pierce, R. S. Water Resour. Res. 1969, 5 , 1353-1363. (45) Baumann, E. W. J. Inorg. Nucl. Chem. 1969,31,3155-3162.

Received for review March 19,1986. Revised manuscript received December 16, 1986. Accepted March 2, 1987.

Environ. Sci. Technol., Vol. 21, No. 6, 1987

601