1960
Anal. Chem. 1980, 52, 1960-1962
Exchange of Comments on Scheme for Classification of Heavy Metal Species in Natural Waters Sir: Several recent publications have delineated a general scheme for determining the distributions of various metals in natural waters among ionic, complexed, and colloidally associated forms (1-5). These publications have also presented data regarding the significance of these various chemical forms (species) in selected natural water systems. The general scheme operationally defines four general classes of metal species as described in Table I. Class 1 includes free metal ions plus organic and inorganic complexes which can be reduced in the time scale of the anodic stripping voltammetric (ASV) measurement used. Class 2 includes ASV-active metals sorbed on organic and inorganic colloids which pass through a chelating resin column. Class 3 includes organic or inorganic metal complexes which are not reduced during the deposition step in ASV but which are dissociated by a chelating resin column. Class 4 comprises ASV-inactive metal associated with organic and inorganic colloids. The overall scheme employs eight analytical steps following four general treatment steps to determine the respective species concentration. Details of the procedures used can be found in the literature ( 1 , 5 ) . Florence and Batley have made, through this scheme, an important contribution to understanding the complicated problem of characterizing metal species in natural waters. Their approach will find widespread application to a problem of longstanding importance. We are in general agreement with the utility and appropriateness of operational definitions in this area of research. However, in order to satisfy fundamental criteria for intercomparison of results, the operations must be defined precisely and the sensitivity of the results to variation of operational parameters must be well understood. Although Florence and Batley stressed that their scheme for differentiating between species is based on behavioral differences (1-5), it is our contention that this must be emphasized, that the meanings of the commonly used terms “labile” and “nonlabile” must be carefully delineated and that the operational protocol requires careful consideration in order to infer what effects the choices of chemical and/or instrumental parameters may have on the speciation results. These contentions serve as foci for the present communication. As Crow (6) has pointed out, considerable confusion arises between the terms lability and stability. In the present context, lability refers to the ease with which substitution or displacement processes occur. Thus, it implies the kinetic stability of a particular species. The thermodynamic stability of a complex, for example, is expressed quantitatively in terms of a stability constant whose significance and determination are well-known. Thermodynamic stability indicates the extent to which a complex will be formed from, or be transformed into, other species in an equilibrium system. Kinetic stability indicates the rate a t which changes will occur which lead to the attainment of equilibrium. Thus, if the kinetic stability is low enough that chemical or physical treatments of the sample lead to the attainment of a new status of thermodynamic equilibrium before or during the analytical measurement, the species being converted to that status by the treatment may be classified as labile. Those which do not achieve that equilibrium status within the time scale of the treatment and/or the measurement are categorized as nonlabile. In the former situation, thermodynamic stability is the parameter of prevalence; in the latter case the kinetic stability becomes the predominent influence on the measurement results. Indeed, the time lapse between the sample treatment and the analysis steps and the duration of the 0003-2700/80/0352-1960$01 .OO/O
analytical measurement itself are both likely to affect the ability to discriminate between labile and nonlabile species when kinetic stability is important. Judging the extent of importance of the kinetic limitations is clearly contingent on the characteristics of the species involved as well as the chemical and physical manipulations to which the samples are subjected. Although the use of the full speciation scheme requires 2 man-days of effort ( 5 ) ,the kinetic limitations operative at each stage of the analysis will depend on the time elapsed between, or during, each treatment step and the actual measurement steps. For steps requiring a few minutes for completion, equilibrium may not be reached. Therefore, inferring the possible effects of sample treatment steps on species distributions via thermodynamic evaluations assuming equilibrium is likely to result in worst case estimations. This possibility should be borne in mind in considering the individual cases discussed below. Effects of p H Adjustment. The ASV measurements are carried out on samples made 0.016 F in acetate buffer at p H 4.8. It is well-known that the nature and extent of complex formation often depend on pH. The importance of inorganic species such as carbonate and hydroxide in forming soluble and insoluble substances with several metals is well recognized. Further, the degree of ionization and the associated ability of many organics to form metal-ligand systems are often affected by pH. Therefore, the adjustment of the p H of the water sample from its native value to that of the acetate buffer (pH 4.8) may result in either increased or decreased ionization and metal binding. Thus, the definition of the term “labile” for all ASV measurements made without HNO, digestion should be qualified to include pH, as well as ASV, lability. Since a large fraction of natural water samples exhibit p H levels above 5-6 and many potential ligands have pK values greater than 5, the effect of pH adjustment on shifting the relevant complexation equilibria may be very prominent. T o illustrate this, we analyzed a K N 0 3 solution (0.1 F) containing nominally F concentrations of Cd, Cu, and P b by ASV for the electrochemically labile concentrations of each element a t pH 6.8, as an approximation of a “natural” water sample, with and without fulvic acid present a t 1 X F. The addition of fulvic acid was predicated on the basis of its rather ubiquitous occurrence in nature and its potential prominence as a ligand for metals (7). Subsequently, the pH of both solutions was adjusted to 4.8 by HNO, addition and the measurement repeated. The results summarized in Table I1 clearly illustrate that pH adjustment causes changes in the apparent fractions of the metals bound by fulvic acid. These changes may be representative of equilibrium states a t the respective pH levels or they may be due to dissociation processes which occur at different rates for each p H during the measurement step and they reflect the presence of some metal-fulvic acid complexes which are electroactive ( 4 9 ) . In either case, the important observation is that the results demonstrate a sensitivity to pH adjustment. While this is only one example, it illustrates the fact that pH adjustment should be avoided when speciation measurements are made if it is possible. Effects of Ligand Competition. Although the stability constants for the complexation of metals by acetate ion often have values less than IO4, it must be recognized that the concentration of the acetate buffer added may be sufficiently high to force displacement of other ligands by the acetate ion. To illustrate the potential significance of this, consider a water IC 1980 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 52, NO. 12, OCTOBER 1980
1961
Table I. Outline of the Florence and Batley Scheme for Metal Speciation ( 1 ) species measured after indicated analytical treatment ASV analysis in acetate buffer after HNO, ASV analysis in acetate buffer to digestion to get total of step all metal forms get “labile” species effect of treatment treatment no. total of all metal forms free metal ions ( M ) plus “electroremoves particulate 1 filter through 0.45-pm filter chemically labile” organic complexes (MLl), inorganic complexes (MAl), and inorganic colloidal species (MA2) metals adsorbed on organic colML42 + MA2 plus colloidal species 2 aliquot passed through metals included in orpassed; free and loidal species (ML2) and inorChelex-100 column ganic (ML4) and inordissocia ble ganic colloidal species (MA2) equilibrated with Na’ ganic (MA4) colloids species retained total of all metal forms releases metals M + M L 1 + MA1 + ML2 + MA2 aliquot irradiated with UV 3 plus “strong” metsl-organic light plus addition of H,O (ML3) and metal-inorganic (MA3) ligand systems MA2 + MA4 MA2 4 aliquot passed through free, dissociable, Chelex-100 column after and UV released UV irradiation species retained Table 11. Apparent Percent of Total Metal Bound by Fulvic Acida 5% of total metal bound by fulvic acid Cd cu Pb
49.5 92.6 47.3 pH 6.8 9.9 80.9 1.4 pH 4 . 8 5% metal released 39.6 14.7 45.9 by pH ndjustment a Determined by differential pulse measurements using a hanging mercury drop electrode. sample having a total copper concentration of lo4 F, a ligand (L) concentration of F, and a constant for the complex CuL of lO+’O It’. If no other ligands capable of binding copper are present, equilibrium computations indicate that essentially all copper would be present as CuL and the free ligand concentration would be -9 x F. Addition of the acetate buffer to this sample according to the recommended conditions ( 1 ) would provide a free acetate concentration of 8 X F at this pH t o compete with the ligand. Acetate (Ac) reacts with Cu2+on a stepwise basis to form CuAc2having successive formation constants of 1.4 X lo2 and 1.1 X lo1 ( I O ) ; neither is particularly large. On the assumption that the p H adjustment associated with the acetate addition has no effect on the displacement of Cu from CuL and that the ligand competition reaction has time to attain equilibrium, the influence of the acetate can be estimated by standard computational procedures (11). Calculations for this example show that the concentration of copper in the forms CuAc and CuAc2 F or about 50% of that originally bound would be - 5 x by the ligand L. This example suggests that acetate addition can sharply shift metal-ligand equilibria even when the formation constants for the competing ligand systems are considerably higher than those of the metal-acetate systems. An example of the possible significance of this was obtained by comparison of the free metal concentrations in the metal-fulvic acid solutions described above with both K N 0 3 (0.1 F) and an acetate buffer at 0.016 F as supporting electrolytes at p H 4.8. The presence of the acetate resulted in shifts in the metal-fulvic acid equilibria producing 8.5, 14.4, and 22.3% more electroactive Cd, Cu, and Pb, respectively. Again, these shifts in the apparent concentrations of the aquo ions may not have been explicitly representative of an equilibrium status, but they are indicative of the possible significance of
ligand competition effects associated with the addition of acetate. Mercury Competition Effects. Mercuric ion tends t o form complexes with numerous ligands and these often exhibit relatively large formation constants. The addition of Hg2+ at F to allow continuous replenishment of the thin-film electrode by electrodeposition can consequently cause the displacement of several metals from their various bound forms. Since the formation constant for the mercury acetate complex (HgAc2)is about 3 X los F-* (10)the Hg(I1) may be essentially all bound in this form. The competition between Hg(I1) and other metal ions for the other ligands which may be present will Consequently involve competition between these ligands and the acetate as well. Again, a hypothetical example may be used to estimate the potential seriousness of such effects. Given a ligand (L) present a t a free concentration of lo4 F and setting the criterion that the HgL concentration must be equivalent to 10% of the total Hg concentration, i.e., 1 x lo4 F, to be significant, calculations (11) suggest that a formation constant for the HgL species exceeding approximately 10l1 F-’ would be necessary to cause a shift in equilibrium of this magnitude. T o some extent, acetate may serve as a masking agent for Hg(1I). Model computations, assuming the presence of another metal at F and applying the criterion that Hg would have to displace 10% of the metal from its ligand to cause a significant shift, suggest that the HgL formation constant would have to exceed those of the other metd-ligand systems by lo3 or more. Although this requirement would change for other metal-to-ligand reaction ratios, these results indicate that displacement of other metals from their ligands by the added Hg is possible. Again, the extents of such reactions would depend on the kinetic and the thermodynamic features of the system studied. Comparisons of the electroactive concentrations for the metal-fulvic acid solutions described above were made at pH 4.8 and 6.8 by using Hg drop and the in situ deposition of F Hg(I1). The in situ deposition a t pH 4.8 resulted in 12.1% increases in the apparent aquo ion concentrations of Cd and Cu but showed an apparent decrease of 45.2% for the electroactive lead concentration. At pH 6.8, the apparent amount of electroactive Cd was decreased by 44%; the stripping current for P b was increased by 51.570, and there was no change in the copper. Clearly, such observations verify the complexity as well as the potential seriousness of the Hg(I1) competition problem. It must be stressed, however, that fulvic
-
1962
Anal. Chem. 1980, 52, 1962-1963
acid tends to adsorb onto the Hg electrode and causes shifts in the apparent concentrations of aquo and complexed forms of the metals (8, 9). The extents of such effects will clearly change between the preformed and the in situ modes of operation, The present experiments have been insufficient to delineate the significance of the Hg competition effect relative to the fulvic acid adsorption interference effect. Therefore, the above data may reflect both effects. Although the value of the computational evaluations presented above depends largely on a lack of kinetic stability for the species present in natural water systems, they have been used as vehicles for demonstrating the potential influences of the recommended analysis conditions when the species are kinetically labile. Similarly, the limited experimental results presented are not necessarily representative of the attainment of equilibrium in each instance. They have been included primarily as illustrative examples of shifts in the estimation of electroactive ion concentrations which may be induced by various stages of the recommended sample treatment protocol. T h e computations and the example results indicate that the effects considered may, individually or collectively, play prominent roles in shifting the native equilibria extant in water samples into entirely nonrepresentative conditions. Florence and Batley have emphasized that speciation by their procedures relies on behavioral differences and that the use of the term "labile metal" requires a statement of the exact experimental conditions used. They have also stressed that the fraction of metal contributing to the ASV stripping current may be comprised of electroactive complexes as well as aquo ions. The present authors strongly support these emphases; such speciation measurements must be qualified within the
RECEIVED for review December 21, 1979. Accepted June 26, 1980. Supported in part by EPA Grant No. R805-183-02.
Sir: Skogerboe et al. quite correctly point out that the trace metal speciation scheme described by Batley and Florence (1,2) may not accurately measure the concentrations of the various species originally present in a water sample, because (a) the pH of the sample is changed before measurement, (b) acetate ion is added, and (c) mercury(I1) nitrate is added for in situ mercury film formation. We were, in fact, very careful to emphasize in our applications (2-4) of this speciation scheme that the results obtained are operationally defined and do not necessarily reflect the true equilibrium concentrations of metal species. However, it should not be assumed that, by omitting acetate and Hg(I1) and maintaining the pH at the natural value, the scheme will yield the original equilibrium species concentrations. Anodic stripping voltammetry (ASV) is a dynamic method of analysis, and all labile metal complexes will dissociate to some extent at the electrode surface ( 5 ) ,whatever the pH. Other perturbations from equilibrium may take place during passage of the solution through the chelating resin column and as a result of UV irradiation. Even the initial filtration of the sample through a membrane filter may affect the speciation by removing adsorbent in the form of particulate matter and by changing the concentrations of dissolved oxygen and carbon dioxide. T h e measurement of ASV-labile metal at the natural pH of the water, and in the absence of buffer, poses some practical problems. The slopes of the ASV peak height-metal concentration curves for Cu, Pb, Cd, and Zn are sensitive to pH and decrease rapidly a t pH values higher than 6 (6). Each water sample would have to be carefully calibrated by standard addition of metal, a procedure difficult to perform without altering the p H of unbuffered waters of p H around 7 . Deaeration of unbuffered waters also alters their pH, and electrolysis of unbuffered, near-neutral solutions results in the p H a t the electrode surface varying considerably from that of the bulk solution. With many fresh waters, an inert elec-
trolyte would have to be added to provide sufficient electrical conductivity, particularly if differential pulse ASV is used. ASV measurements on unbuffered waters would therefore be poorly characterized by comparison with measurements made in a buffered solution of constant pH. ASV a t a thin mercury film electrode (TMFE), formed in situ by simultaneous deposition of mercury and trace metals, is now widely recognized as the most convenient and sensitive ASV technique, particularly when the differential pulse mode is used (7-10). When we first applied the in situ mercury deposition method to trace metal speciation, we carried out experiments on seawater and freshwater in which results with a T M F E were compared when the electrode was both preformed and in situ formed. No significant differences were found. This result can be explained by assuming that if a complex, ML, is sufficiently labile to exchange with M Hg(I1) after about 15 min a t 25 "C, then it will also exhibit considerable ASV lability, and the concentration of M measured in the absence and presence of Hg(I1) may not be greatly different. A similar argument could be used to predict that low concentrations of acetate would not seriously affect labile metal measurements. The very large differences found by Skogerboe et al. between preformed and in situ deposited films are surprising, and we wonder if they were the result of calibration problems caused by working with unbuffered solutions. Also, high concentrations of fulvic acid lead to electrode fouling, and calibration by standard addition under these conditions is difficult because the calibration curves are nonlinear. It is quite possible, however, that in some samples the addition of Hg(I1) and acetate could affect speciation results. This is not a problem as long as it is understood that the operationally defined scheme includes the addition of fixed concentrations of Hg(I1) and acetate to the sample. Since it is impossible to implement a speciation scheme involving ASV
0003-2700/80/0352-1962$01 .OO/O
context of the equilibrium shift possibilities which may be induced by the sample treatment and measurement steps. Ideally, they should also be referenced to the results of similar measurements on well-defined metal-ligand systems as validation models. Without such model evaluations, the actual analytical efficacies of speciation methods cannot be reliably assessed nor can the results obtained be realistically evaluated in terms of potential significance.
LITERATURE CITED (1) (2) (3) (4) (5) (6)
(7) (8) (9) (10) (11)
Batley, G. E.; Florence, T. M. Anal. Lett. 1976, 9 , 379-388. Batley, G. E.; Florence, T. M. Mar. Chem. 1976, 4 , 347-352. Batley, G. E.; Gardner, D. Estuarine Coastal M a r . Sci. 1978, 7, 59-66. Florence, T. M.; Batley. G. E. Estuarine Coastal Mar. Scl. 1977, 75, 79 1-797. Florence, T. M. Water Res. 1977, 1 7 , 681-687. Crow, D. R. "Polarography of Metal Complexes", Academic Press: London, 1969, pp 2-10. Schnitzer, M.; Skinner, S. I. M. Soil Sci. 1966, 102, 361-365. Greter, F. L.; Buffle, J.; Haerdi, W. J . Electroanal. Chem. 1979, 101, 21 1-229. Buffie, J.; Greter, F. L. J . Electroanal. Chem. 1979, 231-251. Skoog, D. A,; West, D. M. "Fundamentals of Analytical Chemistry", 3rd ed.; Holt, Rinehart, and Winston: New York, 1976; p 786. Reference 10, p 280.
'
Present address: Department of Chemistry State University of New York, Buffalo, NY 14214.
R. K . Skogerboe* S. A. Wilson J. G. Osteryoung' Department of Chemistry Colorado State University Fort Collins, Colorado 80523
:E 1980 American Chemical Society
