Anal. Chem. 1@8@, 61 483-488 I
483
Characterization of Metal Binding Sites in Fulvic Acids by Lanthanide Ion Probe Spectroscopy J. C. Dobbs, W. Susetyo, F. E. Knight, M. A. Castles, and L. A. Carreira* Department of Chemistry, University of Georgia, Athens, Georgia 30602
L. V. Azarraga Environmental Research Laboratory,
US.Environmental Protection Agency, Athens, Georgia 30613
Naturally occurring humic substances are known to be potentially strong binders of metals in the environment. A sensitive spectroscopic technique, based on the unique luminescent properties of the triposithre lanthanide metal ions, has been developed to seiectlveiy probe metal blnding sites in humic substances. A continuous multiple site ligand model is proposed to describe complexation of metals with humic materials in terms of mean Mnding strengths, distributlons, and concentrations. Experimental resuits as well as a simulation study that uses this model are also presented.
INTRODUCTION Humic substances are present in all terrestrial and aquatic environments. These materials are complex in nature and vary in composition depending on their geographical location. It is generally assumed that humic substances contain a t least two classes of acidic binding sites, carboxylic and phenolic functional groups. These functional groups are known to bind strongly with metal ions in the natural environment. The biological and physicochemical properties of metals are often changed dramatically as a result of complexation with humic materials (1). Complexation of metal ions can often alter the toxicity of metallic elements. Metal ion complexation can also provide a source of rapid and efficient geochemical transport of metals throughout the environment. In order to develop a site-dependent model for metal complexation with humic materials, a detailed knowledge of the material is required and a method for directly probing these acidic binding sites is necessary. This probe not only should assess the total metal bound but also should be sensitive to the bound ion environment, i.e. the distribution, geometry, and the binding strengths of the different sites. Both discrete multiligand models and continuous multiligand models have been used to describe the binding of protons and metals by humic substances. An excellent critical review of many of these models has been discussed by Perdue and Lytle (2). In this paper, we have described the development of a sensitive spectroscopic technique to selectively probe the metal binding sites in humic substances. This technique is based on the unique luminescent properties of the tripositive lanthanide metal ions, Ln(II1). The use of lanthanide metals as fluorescent probes has been well documented (3-13). The Eu(II1) metal ion has been the lanthanide of choice in most of these applications and its spectral features are widely documented. We have also described a continuous multiligand model to characterize the metal binding sites of humic materials based on the results of the lanthanide ion probe. A single mode distribution model has been presented. It appears to adequately describe the data a t low pH conditions, where the ionization of the more weakly acidic sites was negligible. In 0003-2700/89/0361-0483$0 1.50/0
the present work, only the unprotonated carboxylic sites were assumed to be available for binding with metal ions.
EXPERIMENTAL SECTION Lanthanide Ion Probes.Lanthanide ions and their complexes are of little environmental importance; however, the unique spectral characteristics of Ln(II1) ions render them useful as spectroscopic probes. The environmentally more interesting d-metal complexes and alkaline-earth metals lack the necessary spectroscopic properties to serve as useful probes. The fluorescence spectrum of the Eu(I11) ion contains information about the binding environment of the complexing ligand. The Ln(1II) ions have their lowest lying excited states comprised of various 4fn configurations. The 4f orbitals are largely shielded from the surrounding environment by the outer lying d-electron orbitals and are minimally involved in bonding. The total amount of ligand-field splitting of an f-electron term is small, rarely exceeding a few hundred wavenumbers. Consequently, radiationless deexcitation processes in Ln(II1) ions are relatively inefficient and the emission of radiation as luminescence is able to compete in many instances (14). The transition states between the 4P configurations are formally electric dipole forbidden. Consequently, the absorption and emission of energy between these states are weak. A very small molar extinction coefficient, approximately 10 M-' cm-', results in low transition probabilities and long luminescence lifetimes. Lifetimes of several milliseconds are experimentally observable. In contrast, the humic organic materials have molar extinction coefficients that can range in the tens of thousands and fluorescence lifetimes that are typically in the nanosecond range (14). The disparity in fluorescence lifetimes means that the relatively weak lanthanide ion luminescence can be temporally resolved from the strong background fluorescence of the humic material by means of a fairly simple time-resolved gated detector response. Instrumentation. A Lambda Physik EMG 102 XeCl excimer laser and a FL3002 tunable dye laser were used in tandem to supply a high-intensity tunable source. The averaged output of the XeCl excimer laser was 1.6 W (10 Hz rep rate) at 308 nm. A tunable ultraviolet dye wag used to provide an excitation source of 394 nm, which corresponds to a resonant absorption transition of the Eu(I1) metal ion. A block diagram of the experimental setup is illustrated in Figure 1. The beam was focused with a 200-mm lens (Ll) and directed vertically through a standard 1-cm quartz cuvette and the fluorescence emission was collected 90' off axis to minimize stray radiation. Lenses L2 and L3 were used to pickup and collimate the fluorescent image and focus it onto the monochromator slits. A Jarrell Ash 1-m double focusing monochromator was used to disperse the fluorescence radiation of the Eu(1II) ion. A Products For Research gated socket and an RCA C134A photomultiplier tube (PMT)were used to reject the prompt background fluorescence of the humic material and detect the long lived fluorescence of the lanthanide metal ion. A Stanford Research Systems Model 535 digital delay pulse generator provided the necessary timing control for the experiment. The timing sequence generated by the digital delay pulse generator is illustrated in Figure 2. This device was triggered by the excimer laser. It subsequently set the trigger To(85 ns 0 1989 American Chemical Society
ANALYTICAL CHEMISTRY, VOC. 61, NO. 5, MARCH 1, 1989
484
Car C a r B u s Generator
Computer
Supply
a
Dl*PlF'Y
EMG
FL 8002 Dye Laser
102
Excimer Laser
I
Power L1
. ,....
SUPPlY
L3 L 2
Flgum 1. Block diaqam of experimental setup and electronics in.
terface.
Timing Sequence 1
I
1
Excimer Laser
--> Box Car 2 n1 AB --> PMT 1 CD --> BOX car 1 CD --> Sample Hold To
1
A
.
B =
C D
-
i
To A
f
200 usec
*
2 0 0 usee
A * 1 1 0 usec
C
+
10 usec
7
Flgure 2. Timing sequences set by the delayed pulse generator.
after the laser pulse) and established time zero for the experiment. A 2Mrs delayed pulse, AB, altered the reverse biased state of the PMT, and the tube was turned "on" for 200 ra (width of AB). The 200 ,a delay allowed for complete rejection of the shortlived background fluorescence. One hundred ten microseconds into pulse AB, a 10-,a pulse, CD, set the trigger for a sample and hold amplifier which sampled the resulting voltage from a current to voltage converter directly off of the PMT. The conditioned signal was then sampled by boxcar 1 (Model 510 Stanford Research Systems) with an fK+@ delay to allow time for the signal to settle. The output of the boxcar was directed into the IEEE interface (Model SR 245 of Stanford Research Systems) equipped with a 12-hit analog to digital (A/D) converter. A beam splitter (Bl) was used to reflect a small fraction of the dye laser output onto a photodiode detector. This reference signal monitored changes in power of the dye laser as a function of time or wavelength. The output of the photodiode was directed into boxcar 2. The result.1 of each conversion (from boxcar 1 and boxcar 2) were stored in the memory of the interface. A t the completion of the scan, the interface was polled and the entire set of data points were read into a file and stored on hard disk. AU meaaurement.1were taken 88 the ratio of the fluorescence signal and the reference signal. Reagents and Sample Preparation. The humic suhstances selected for this study were aquatic fulvic acid standards of the Intemational Humic Substances Society. These standards were isolated from the Suwmee River near Fargo, GA, from November 1982 to February 1983, by Leenheer, Wilson, and Malcolm (25). Stock solutions of the fulvic acids and the Eu(II1) metal ion were prepared. A 6.5-mg portion of the dry fulvic material was weighed and disolved in 50 mL of distilled water, to give 130 ppm of the fulvic acid. The pH of this solution was adjusted to 3.5. This fulvic stock solution was diluted hy I/, to give an 18.5 ppm M EuCI,.GH,O was solution. A 100-mL solution of 2 X prepared. Eleven serial dilutions of the Eu(III) solution were made to give a range of concentrations extending over 5 orders of magnitude. The pH of each solution was adjusted to 3.5. Twelve samples were prepared by mixing 1:1(volume/volume) solutions of the diluted fulvic acids and the Eu(II1) dilutions. AU final solutions were adjusted to pH 3.5. Table I summarizes the initial and final concentrations of each sample. It was determined that complexation of the Eu(1II) metal ion occurred rapidly, and
O.BBI/*iS5
08
Flgure 3. Emission spectra of ENlIl) bound wnh hrMc acid nmterlak (a) 1 X IO3 M Eu(III), 100 ppm fulvic acid; (b) 1 X IOa M Eu(III), 100 ppm fulvii acid; (c) 1 X IO-' M Eu(lI1). 100 ppm fubk acid.
spectmscopicmeasurements could be taken immediately following sample preparation.
RESULTS AND DISCUSSION The most fundamental and necessary requirement of any probe used in metal-speciation studies is the ability to quantitatively determine the concentration of the total bound and free metal in the solution mixture. The lanthanide ion probe can readily measure the concentration of the complexed
ANALYTICAL CHEMISTRY, VOL. 61, NO. 5, MARCH 1, 1989
Ratio Plot
Table I. Experimental Concentrations, pH Conditions, and Observed Ratio Value (ZIz/Z61(l) for Each Sample initial Eu(II1) initial Fa, concn, M PPm 2.00 x 10-2 1.00 x 10-2 4.00 x 10-3 2.00 x 10-3 LOO x 10-3 4.00 X lo4 2.00 x 10-4
1.00 x 10-4 4.00 X 10“ 2.00 x 10”
1.00 x 10” 5.00 X 10”
20 20 20 20 20 20 20 20
30 20
20 20
RATIO
final Bu(1II) concn, M
final Fa, ppm
ratio
1.00 x 10-2 2.00 x 10-3 2.00 x 10-3 1.00 x 10-3 5.00 X lo4 2.00 x 10-4 1.00 x 10-4 5.00 X 2.00 x 10” 1.00 x 10-6 5.00 X 10” 2.50 X 10”
10 10 10 10 10 10 10 10 10 10 10 10
2.65 2.62 2.47
‘616
T w o limiting conditions: is high; CM >> CL 1) CM / C 2) K e f f
2.31 2.02 1.70 1.50 1.21
-
>>
CL
CM
-->
-->
Xb
XS
I
Xb
0.56 0.77
0.58 0.55
metal and the aquo or free metal ion. This information is encoded in the experimental Eu(II1) fluorescence emission profiles. Shown in Figure 3 are the fluorescence spectra of three Eu(1II) and fulvic acid mixtures. Each sample contained equal amounts of fulvic acid, but different concentrations of total Eu(III), lo*, and lo-’ M, respectively. These spectra are obviously very different. The differences in the ratio of the intensities of the two emission bands in these spectra were due to the hypersensitive nature of one of the transitions. Hypersensitive Transitions. The basis of the experimental determination of the binding of lanthanide ions with humics lies in the existence of the hypersensitive emissive transition of the Eu(II1) ion, 5Do 7F2(616 nm). Hypersensitive transitions are specific absorption or emission transitions of the lanthanide metal ions that are extremely sensitive to ligation. The intensity of a hypersensitive transition is enhanced in the metal-ligand complex relative to that transition in the aquo ion. Experimentally, this was observed by making the concentration of the Eu(II1) small relative to the amount of ligand present such that the majority of the metal ions were complexed with the ligand. In this treatment, we assumed that the hypersensitive effect was the same for all ligand sites. Since the ligand of interest at the pH values studied was the carboxylate ion, this assumption seems justified. The hypersensitive effect has been treated theoretically by several researchers (16) and is generally thought to be the result of a quadrupolar interaction of the ligand with the metal. For known hypersensitive transitions, most ligands are expected to produce the hypersensitive effect with respect to the aquo ions; however, there may be some exceptions. The degree of signal enhancement in the hypersensitive transitions depends on the specific system. Intensity enhancements as large as 100-fold have been reported for the 5D0 7F2hypersensitive transition of Eu(II1) ion (17). For the case of the fulvic acid ligand, the hypersensitive transition of the Eu(II1) ion, 5Do 7F2(616 nm) was observed, and its effect is clearly shown in Figure 3. The intensities of the 592-nm and the 616-nm peaks become inverted as a function of CM,the total concentration of the metal added. Because of the large CM value relative to the total ligand concentration, Figure 3a is essentially the fluorescence spectrum characteristic of the Eu(II1) aquo ion. The value of CM in Figure 3c is small relative to the total ligand concentration and therefore is more representative of the fluorescent emission spectrum of the complexed metal. The enhancement of the 616-nm peak is clearly shown. The hypersensitive effect is a continuous function of the metal-ligand complexation and therefore provides a direct quantitative measurement of the state of the metal ion throughout the entire titration process. Ratio Plots. The result of plotting the ratio of the intensity of the nonhypersensitive transition a t 592 nm to the hyper-
-
-
485
-
-6
iog Cm
-‘
Flgure 4. Representation of a ratio plot curve. The ratio values asymptotically approach X , when the majority of metal added is bound to the humic ligand.
sensitive transition a t 616 nm as a function of CM yields a sigmoidal-shaped curve. This sigmoidal curve can be treated as a “spectral” titration curve. A representation of this curve is depicted in Figure 4. This sigmoidal curve contains the necessary information to calculate the concentrations of the bound and free metal species. The analytical expressions for Z[MLi] and [MI were derived from a mathematical treatment of the spectral titration curve. Let the relative ratio of the ntensity of these two emissive peaks be defined as
R = I592/I616
(1)
Figure 4 depicts the two extreme limiting values of the ratio plot shown. These conditions represent the maximum and minimum ratio values, x b and X,. When the value of c M / c L is large (greater than loo), where CL is the total concentration of the humic ligand, the majority of the metal in solution is free and is representative of the limiting condition Xb Correspondingly, when the vlaue of CM/CJci, where Ki is the effective binding constant, is small, the majority of the metal is bound with the ligand and is representative of the limiting condition X,.Both of these limiting conditions were experimentally obtainable. The intensities of the emission transitions a t 592 and 616 nm depend on the concentrations of both the bound and free metal species and their respective contribution factors
+ Z[MLi]X
(2)
16160: [MI Y + 2[MLi] YT
(3)
1592
E
[MIX
Let X be the fluorescence contribution factor for the nonhypersensitive band occurring a t 592 nm. This factor is the same for both of the metal terms since the intensity of this band is not influenced by the metal-ligand complex. The emissive transition occurring a t 616 nm is a hypersensitive band. The corresponding intensity contribution factors, Y and Y T , are not equal since the complexed metal term, 2[MLi], is influenced by the hypersensitive effect. From eq 1-3, the ratio of the intensity, R, can be expressed as [MIX + Z[MLi]X R= (4) [M]Y + B[MLi]YT where T is a constant greater than one. For the limiting condition when CM/CLis large, i.e. 100 or larger, 2[MLiJ
