Anal. Chem. 1989, 6 1 , 525-529
525
Automated Two-Column Ion Exchange System for Determination of the Speciation of Trace Metals in Natural Waters Yan Liu’ and J. D. Ingle, Jr.*
Department of Chemistry, Oregon State University, Gilbert Hall 153, Corvallis, Oregon 97331 -4003
A rapid method for determination of metal speciation based on an automated two-column ion exchange system is described. Two fractions of dissolved trace metal species are directly determlned by on-line flame atomic absorption spectrophotometry after preconcentration by sequential columns of Cheiex-100 chelating resin and AG MP-I macroporous anion resin. A third fraction Is determined by standard addition. Variables that affect the results obtained by the twocolumn system are studied by the use of model compiexlng agents. With a IOmL sample loop, the sample throughput is 6 samples per hour and detection limits are 0.1 bg/L for Cu( II), 0.08 bg/L for Cd( II), and 0.2 pg/L for Zn( 11). The method is used to determlne the speciatbn of Cu( II ) , Cd( I I ) , and Zn(I1) in natural water samples.
Knowledge of chemical speciation of trace metals in natural waters is essential for the interpretation of biological or geochemical cycling of trace metals in natural waters. In recent years, several measurement schemes have been reported for the determination of chemical speciation of trace metals in seawater and freshwater (1-7). Most involve a series of sample treatment, separation, and measurement steps and are rather complex and time-consuming (up to 8 h per sample). In addition, the classification of metal species provided by many schemes is often more detailed than that required to define the biological effects of interest such as toxicity of trace metals to aquatic organisms (8,9). Relatively simple and rapid measurement schemes are needed to establish the correlation between metal speciation and biological effects. Chelex-100 resin has been widely used in the determination of chemical speciation of trace metals in natural waters and is a key step in most measurement schemes. Typically, a filtered (0.4 pm) water sample is stirred with Chelex-100 resin in a beaker (batch mode) or is passed through a column containing Chelex-100 resin (column mode). The “Chelex labile” fraction is that retained by the resin and includes hydrated metal ions and weakly complexed or bound metal species that dissociate. This fraction is determined with techniques such as atomic spectrometry after stripping the retained metal, typically with acid. The “Chelex nonlabile” fraction is that not retained by the resin and likely includes stable metal complexes with fulvic and/or humic acids and metals strongly associated with (i.e., complexed by or adsorbed on or occluded in) colloidal particles. This fraction can be determined by difference if the total metal concentration is known. In all the measurement schemes reported (l-n,Chelex-100 separation is carried out in an off-line batch or column mode. Recently, several on-line ion exchange preconcentration system using Chelex-100 resin have been reported in which the retained metal species are eluted directly into the nebulizer of *Present address: Dionex Corp., 1228 Titan Way, Sunnyvale, CA
94088.
an atomic spectrometer (see ref 10 and the references within). These systems yield a given preconcentration fador with much smaller samples and higher sample throughput rates compared to off-line systems. This paper reports an automated two-column ion exchange system that is based on sequential Chelex-100 resin and AG MP-1 resin columns that retain different physicochemical forms of soluble trace metal species. The rapid measurement scheme for trace metal speciation is characterized by using model complexing agents and is used to determine the speciation of Cu(II), Cd(II), and Zn(I1) in natural water samples.
EXPERIMENTAL SECTION Reagents, Solutions, Columns, and Samples. All reagent, complexing agent, and metal ion test solutions were prepared with analytical grade chemicals (unless otherwise specified) and double-deionized water under a class-100 laminar flow hood. A humic acid solution was prepared by dissolving 100 mg of humic acid (sodium salt) from Aldrich Chemical Co. in 1L of deionized water, then filtering the solution through an acid-cleaned 0.4 pm Nucleopore filter. The carrier buffer solution was a solution of 5 mM NH4Ac/l mM HACat pH 5.4. The stripping reagents (SR) were 0.025 M cysteine/0.5 M NH40H (NH,.H20)/2.0 M “,NO3 for Cu(II), 0.1 M EDTA/O.5 M NH40H/2.0 M NH4N03for Cd(II), and 2.0 M HNO, for Zn(I1). The resin regeneration reagent was 2 M NH40H. Chelex-100 and AG MP-1 resins (both 1OC-200 mesh, Bio-Rad Laboratories) were converted into the NH4+ and OH- forms, respectively, with 2.0 M NH40H and as described previously (IO, 11). Altex microbore glass columns (3 mm i.d. X 50 mm) modified as reported previously (10)were packed with the water-slurries of the resins. Water samples from the Willamette River and a rural drainage ditch near Corvallis, OR, were collected and filtered as previously described (10). The water samples were buffered at pH 6.8 in 0.01 M NH,Ac by addition of a small volume of a concentrated buffer. A portion of the filtered and buffered water sample was spiked with Cu(II), Cd(II),and Zn(I1) ions to provide a 6.0 pg/L increase in concentration. Sample solutions were allowed to stand for 8- to 1 2 h before measurement. Calibration curve data were obtained with solutions of 2.5,5.0,10, and 20 pg/L of each metal ion at pH 6.8 in 0.01 M NH,Ac. Apparatus. The experimental system is depicted schematically in Figure 1. The major components and construction of the system are basically the same as previously reported for an on-line, one-column trace enrichment system (10). However, to accommodate the second ion exchange column, a second pair of column stripping reagent (SR 11) and regeneration reagent (“,OH) sample loop valves (V8 and V9) are added for eluting and regenerating the AG MP-1 ion exchange column. Two three-way valves (V5 and V7) and one four-way valve (V6) are also added to direct the flow path of the carrier buffer stream. Two dualchannel peristaltic pumps (Per. Pump I1 and Per. Pump 111) are used for loading valves V2, V3, V8, and V9. The atomic absorption (AA) spectrophotometer was operated with an *acetylene flame and the conditions were described previously (10). The operation of the automated two-column system consists of four steps. In step one, sample solution, SR I, SR 11, and NH40H solutions are loaded into their respective sampling loops
0003-2700/89/0361-0525$01.50/00 1989 American Chemical Society
526
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
TIME
experimental runs with the two-column ion exchange system. Sample was 200 pg/L Cu(I1) 8 mg/L humic acid at pH 7.
Figure 2. AA signal profiles of
+
V I
Figure 1. Block diagram of flow and detector components of the automated two-column ion exchange system.
by peristaltic pumps (60 and 20 s for a 10- and a 2-mL sample loop, respectively). The reagent peristaltic pumps are turned off after 8 s to conserve these reagents. In step two, the positions of V1, V2, V3, V8, and V9 are switched simultaneously by a single pneumatic activator to the "inject" position. Carrier buffer (from pump A through V5 and V4 at typically 5 mL/min, pump B is used when a flow rate above 10 mL/min is required) pushes the sample solution plug from V1 through the sequential columns (300 and 80 s for a 10- and a 2-mL sample loop, respectively). In step three, the positions of valves V4, V6, and V7 are switched solutions and the carrier stream now directs the SR I and ",OH in V2 and V3 to the Chelex-100 column to elute the metal species retained on this column. The plug of SR I solution carries metal species through V6 and V7 into the nebulizer of the flame AA spectrophotometerand a transient AA signal produced is recorded. In step four, the positions of V5 and V7 are switched such that the carrier stream now pushes the SR 11and the NKOH solutions in V8 and V9 to elute the metal species retained on the AG MP-1 column into the AA nebulizer and to regenerate the resin. The second elution peak is similarly recorded. The elution and regeneration of the column and complete washing of the flow path in steps 3 and 4 require 60 s each. All the valves are returned to their initial positions and the carrier stream flow is maintained for 90 s to condition the columns before next run. An AIM-65 microcomputer system completely controls the operation of the system including switching all valves, starting and stopping pumps, signal acquisition, and data processing. The software is configured to report the peak height, peak area, and retention time (relative to the time of initiating the stripping step) for the AA peaks resulting from the elution of each column and statistical data for repetitive runs.
RESULTS AND DISCUSSION Two-Column Measurement Scheme. The present measurement scheme separates the dissolved trace metal species in the water sample into three fractions. The Chelex-100 column retains hydrated metal ions and metal ions liberated from metal species that dissociate during passage through the Chelex-100 column. The latter species may include labile metal complexes and possibly metals loosely associated with colloidal matter. The metal species retained by the Chelex-100 column are referred to as M1 species. Nonlabile metal complexes and metals strongly associated with colloidal matter are not retained by the Chelex-100 column. The AG MP-1 resin is a macroporous strongly basic anion exchange resin with a large pore size that can retain metal complexes that are negatively charged and some metal ions associated with negatively charged organic matter such as humic material. These metal species are referred to as M2 species. Metals strongly associated with very large colloidal particles may not be retained because of the molecular exclusion limit of the AG MP-1 resin (molecular weight about 75000 (12)). Also neutral nonlabile metal complexes may not be retained by the column, although retention due to hydrophobic interactions is possible. Unretained metal species make up the third fraction and are referred to as M3 species. The use of two sequential columns to retain two classes of metal species in an off-line fashion has been demonstrated
by several researchers (13-16). Anion exchange resins such as AG 1-X8 (strongly basic) (14), and Sephadex A-25 (weakly basic) (161,have been used to retain anionic metal species or metals associated with negatively charged humic substances. In this study, several anion resins, including Bio-Rad AG 1-X8, Bio-Rex 5 (intermediate basic), and AG MP-1 resins and Sephadex A-25 resin, were tested. Sephadex A-25, a gel material, collapses under the operating pressure of the flow system (typically 60 psi). The AG 1-X8 and Bio-Rex 5 resins are less efficient than the AG MP-1 resin in their retention of the humic complexes of Cu(I1) because they have smaller molecular exclusion limits (e.g., molecular weight about 2000 for the AG 1-X8 resin). The AG MP-1 resin was therefore chosen for all further studies. It is critical that the Chelex-100 column is placed in front of the AG MP-1 column; fresh AG MP-1 resin retains trace amounts of free Cu(I1) ions. Furthermore, the color of the AG MP-1 resin changes from off-white before use to beige after a few runs with sample solutions containing humic acid because some of the humic material is retained and not completely eluted from the resin after each sample run. Some of this permanently retained organic matter is capable of binding trace metal ions. In fact, after the AG MP-1 column was used for a few sample runs with sample solutions containing humic acid, it retained most of the hydrated Cu ions in 1.0 mL of a 200 pg/L Cu2+solution. Complexing agents can be used to elute trace metal ions retained by Chelex-100 resin (IO). A suitable SR solution is also needed to elute trace metal species retained by the AG MP-1 column. A solution of 2 M ",NO3 was found to elute effectively anionic metal complexes such as (CUEDTA)~from this column. However, trace metal species complexed by organic matter such as humic material that is irreversibly adsorbed on the AG MP-1 resin may not be completely eluted off the resin by the nitrate solution. It was found necessary to add cysteine or EDTA to the nitrate solution to strip trace metals from the adsorbed organic matter. The exact compositions of the SR solutions for either column for Cu(I1) and Cd(I1) are specified in the Experimental Section. A solution of 2.0 M HN03 was also effective for elution of metal species retained by the AG MP-1 column and was used as the SR to elute Zn(I1) species retained by the two columns. To avoid the pressure increase problem of the Chelex-100 column when using nitric acid as the SR, two valves (not shown in Figure 1)were added to the system to reverse the flow direction of the carrier buffer stream through both columns during the loading period for the sample loop. The back-flushing technique decreases the compreasion of the resin packing in the column. The SR solution for Cd(I1) can also be used to strip Zn(I1) from the columns if a the stop-flow technique is employed (10). Typical AA signal profiles are shown in Figure 2. Peaks I and I1 correspond to M1 species (Chelex-100 column) and M2 species (AG MP-1 column), respectively. When each column was loaded with an equivalent amount of metal species (i.e., same number of moles of Cu(II)), the elution peak height for the two columns differed by about 15% as shown in Table I. This occurs because the dispersion of the peaks differs due
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
Table I. Peak Height and Area Data for the Elution Peaks from the Chelex-100 Column and the AG MP-1 Column column
peak height"
peak areab
Chelex- looc AG MP-ld
0.183 f 0.001 0.159 f 0.003
5964 f 67 5902 f 51
"The peak height data are in absorbance units (average f standard deviation for four runs). The peak area data are in absorbance units seconds (average f standard deviation for four runs). cSample solution was 200 pg/L Cu(I1) in 0.005 M NH4Ac/0.001 M HAC at pH 5.2; 2-mL sample loop. dSample solution was 200 pg/L Cu(I1) in 0.001 M EDTA at pH 7 ; 2-mL sample loop.
Table 11. Speciation Results of Cu(I1) in Some Complexing Media" solutionb
70 M1
'70 M2
log K'
5 mM NH4Ac/l mM HAC 1.0 mM glycine 1.0 mM IDA 4.0 pM EDTA 2.0 pM EDTA 2.0 pM NTA 8.0 mg/L humic acid
100 f 2 100 1 92f1 ND 36 f 2 42 2 41f2
ND' ND 8*1 100 f 2 64 f 1 58 f 2 52f1
9.9 11.9 15.5 15.5 10.3
*
*
"Results are in percentage (average f standard deviation for three runs), where 70 M1 is the percentage of the copper retained by the Chelex-100 column and % M2 is the percentage of the copper retained by the AG MP-1 column. bThe concentration of Cu(1I) was 3.15 pM in all solutions. All solutions were adjusted to pH 7 except the acetate buffer solution was at pH 5.2. c N D = not detected. log K'is the log of the conditional stability constant of the Cu(I1) complex, adjusted to pH 7 , and an ionic strength of 0.01 M with the value reported (17).
to the differences in the stripping flow paths to the detector, the column packings, the resin particle sizes, and the mechanisms of elution. The peak areas are, however, statistically equivalent. Therefore, the peak area is used to quantitate the amounts of Cu and other metals. The system is calibrated by measuring the peak areas for the Chelex-100 column with standard solutions of free trace metal ions. Alternatively, the AG MP-1 column can be calibrated with a known amount of a negatively charged metal complex (e.g., (CUEDTA)~-) that is completely retained by this column. Speciation of Cu(I1) in Model Complexing Media. Solutions of 200 pg/L (3.15 pM) Cu(II), prepared in different complexing media including EDTA, iminodiacetic acid (IDA), NTA, glycine, and humic acid, were used to characterize the measurement scheme. Speciation results with a 2-mL sample loop are summarized in Table 11. Equilibrium calculations show that 100% of the copper is complexed by the ligand in 1.0 mM glycine, 1.0 mM IDA, and 4.0 pM EDTA solutions, and 64% of the copper is complexed in 2.0 gM EDTA and 2.0 pM NTA solutions. The results in Table I1 demonstrate that all Cu(I1) species in the NH,Ac/HAc and glycine solutions and 92% of the Cu(I1) species in the IDA solution are M1 species. In contrast, the stable Cu(I1)-EDTA complex (i.e., (CUEDTA)~ion a t pH 7 ) does not dissociate during passage through the Chelex-100 column and is a M2 species. All the Cu species are retained by the AG MP-1 column when EDTA is in molar excess relative to Cu(I1) (Le., 4.0 pM EDTA). If the Cu2+ion is in excess (Le., 2.0 pM EDTA), only 64% of the Cu(I1) is a M2 species as predicted by equilibrium calculations. It was also found that the retention of (CUEDTA)~by the AG MP-1 column is independent of the sample loading flow rate up to 15 mL/min. The Cu(I1)-NTA complex is less stable than the Cu(I1)-EDTA complex. A small fraction of the Cu(I1)-NTA complex dissociates during passage through the Chelex-100 column and 42% of t h e Cu(I1) species is a M1
C
1
$
0-l 0.00
527 I
I
0.05
0.10
0.15
0.20
Nitrate Concentration, M Figure 3. Effect of nitrate concentration on the retention of anionic CU-EDTA complex by the AG MP-1 column. Sample was 200 pg/L Cu(I1) in 0.001 M EDTA at pH 7.
species (rather than 36% if there was no dissociation). The results for Cu(I1) species in a solution of 8 mg/L humic acid show that a significant fraction of Cu(I1) is strongly associated with the humic acid and is a M2 species (52%). In addition, the M3 fraction is about 7%. The dissociation of Cu(I1) complexes in the Chelex-100 column depends on the thermodynamic stability and the dissociation kinetics. The values of the conditional stability constants of the Cu(I1) complexes with glycine, IDA, NTA, and EDTA are given in Table 11. The speciation results suggest that the Cu(I1) complexes with conditional stability constants similar to that of the Cu(I1)-EDTA complex (log K' = 15.5 a t pH 7 ) are nonlabile and they do not dissociate when passing through the Chelex-100 column. These complexes are thus measured as M2 species in the present speciation scheme. The Cu(I1) complexes with conditional stability constants smaller or similar to those of the Cu(I1)-glycinate and Cu(I1)-IDA (log K' = 10-12 a t pH 7) are likely to dissociate in the Chelex-100 column provided that the dissociation rate is sufficiently large. These metal species are thus measured as M1 species. The conditional stability constant of Cu(I1)-NTA complex is close to the Cu(I1)-glycinate complex. However, only a small fraction of the Cu(I1)-NTA complex dissociates in the Chelex-100 column possibly due to its slow dissociation. Metal complexes exhibiting slow dissociation are thus measured primarily as M2 species by the present scheme. Because the AG MP-1 resin is a strongly basic anion exchanger, the retention of anionic metal species by the AG MP-1 column may be affected by high concentrations of competing anionic species. The effect of the nitrate concentration on the retention of the anionic Cu(I1)-EDTA complex by this column is shown in Figure 3. The retention efficiency for (CUJ~DTA)~decreases rapidly as the concentration of NO, in the test solution is increased above 0.01 M. The unretained (CUEDTA)~-ion was observed with the on-line flame AA detector during the process of loading sample solution when the nitrate concentration was above 0.05 M. These results suggest that the effect of major matrix anions on the retention of anionic metal species by the AG MP-1 column should be investigated when the present system is applied to high ionic strength samples such as seawater. Freshwater samples should not be a problem since the ionic strength rarely exceeds 0.01 M. The complexation of trace metals by humic substances is very important in their speciation in natural waters. The effect of the solution pH on the speciation of Cu(I1) in humic acid solutions is shown in Figure 4. As the pH value of the Cu(11)-humic solution changes from 5 to 9, the fraction of the M1 species decreases from 60% to 25%, the fraction of the M2 species increases from 36 % to 61 %, and the fraction of
528
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
Table 111. Speciation Results of Cu, Cd, and Zn in Water Samples” sample
M I
Cu(I1) river river + 6.0 pg/L ditch ditch + 6.0 pg/L Cd(I1) river
0.12 f 0.05 1.84 f 0.09 0.29 f 0.07 0.52 f 0.06
+ 6.0 pg/L ditch ditch + 6.0 pg/L river
4.0
5.0
6.0
7.0
8.0
9.0
10.0
Solution pH Flgure 4. Effect of solution pH on the speciation of Cu(I1) in the humic acid solution. Sample was 200 pg/L Cu(I1) in 8 mg/L humic acid.
60 -
-
201 20
-
% Ml
P
/ f
0
0 0
0.69 f 0.06 2.15 f 0.05 2.56 f 0.08 6.95 f 0.28
0.12 f 0.02 4.35 f 0.05 0.13 f 0.02 4.09 f 0.05
NDb
2.69 f 0.23 7.77 f 0.09 1.13 f 0.10 5.90 f 0.12
ND ND
0.24 f 0.06
ND 0.21 f 0.02
ND 0.30
0.08
“Results are in pg/L (average f standard deviation for three runs); 10-mL sample loop. bND = not detected, the detection limits are 0.1 pg/L for Cu, 0.08 pg/L for Cd, and 0.2 pg/L for Zn. The detection limits are calculated from DL = 2sbk/m where sbk is the blank standard deviation and m is the calibration slope.
80 -
40
Zn(I1) river river + 6.0 pg/L ditch ditch + 6.0 pg/L
[M21
5.0
10.0
15 0
20.0
HM Concentration, mg/L
Figure 5. Effect of the concentration of humic acid on of Cu(I1). Sample was 200 pg/L Cu(I1) at pH 7.
the speciation
the M3 species increases from 4% to 14%. As expected, a larger fraction of Cu(I1) is complexed by humic acid at higher pH as there is a lower concentration of H+ ions to compete for binding sites on the humic acid. From pH 5 to 9, the ratio of the M3 fraction to the M2 fraction increases slightly (0.12-0.22). As shown in Figure 5 , the fraction of Cu(I1) complexed by humic acid increases steadily as the concentration of humic acid in the test solution increases. The M3 fraction also initially increases but becomes constant above 4 mg/L HA. The ratio of the M3 fraction to the M2 fraction decreases slightly from 4 to 20 mg/L HA (0.47-0.26). In all measurement schemes involving Chelex-100 resin, the classification of fractions is operational because retention by the resin depends on the contact time of the test solutions with the resin (18, 19). In flow systems, the contact time depends on the sample loading flow rate and the void volume of the column. With longer contact times, the dissociation of moderately labile metal complexes and thus the so-called “Chelex labile” fraction can be greater. The present system is advantageous relative to conventional off-line column techniques. First, the use of a dual-piston pump provides precise control and reproducibility of the flow rate and contact time. Second, the use of small columns and high flow rates decreases the degree of dissociation of moderately labile metal complexes. The contact time is estimated to be about 1.4 s with a 5 mL/min flow rate. For a 200 pg/L Cu(I1) solution in 8 mg/L humic acid at pH 6.5, the percentage of the “Chelex labile” fraction changed from 67% to 49% to 36% when the flow rate was changed from 1 to 5 to 15 mL/min. Application to Environmental Water Samples. The speciation results for Cu(II), Cd(II), and Zn(I1) in water samples presented in Table I11 demonstrate that the present method is applicable to very low metal concentrations. Only a small fraction of the Cu(I1) is the M1 species; it is mostly
the M2 fraction. In contrast, the majority of Cd(I1) and Zn(I1) species are in the M1 fraction; no M2 fraction of Cd(I1) or Zn(I1) species was detected. The speciation results for spiked water samples are also presented in Table 111. The distribution of the added Zn(II), Cd(II), and Cu(I1) between the M1 and M2 metal fractions generally agrees with the distribution in the original sample except for the case of Cu(I1) in the river water sample. Other researchers (2-4) also found that Cd(I1) and Zn(I1) exist primarily as labile metal species and are retained by Chelex-100 resin, while a significant fraction of Cu(I1) is strongly associated with organic matter and not retained by Chelex-100 resin. A limitation of the present system is that the M3 fraction cannot be directly determined. The M3 fraction can be estimated by difference if the total metal concentration can be determined by a technique with sufficient detectability such as electrothermal AA spectrophotometry or by sample digestion followed by preconcentration on the Chelex-100 column in the two-column system. A second method of estimating the M3 fraction is based on a standard addition of a known amount of the trace metal under study. The concentration of the M3 fraction of the added metal is calculated as the difference beween the known total spiked metal concentration and the change in the concentrations of the M1 and M2 fractions measured before and after the spike. If the metal added reaches equilibrium with other components in the water sample and does not alter the original metal speciation, the distribution of the added metal among the three metal fraction an be used to predict the original metal speciation in the water sample. The ratio of the M1 to M2 fraction can be used as a diagnostic tool to test these assumptions. This ratio must be approximately the same for the original and spiked sample to make a valid estimate of the speciation in the original sample. Table IV summarizes the percentage distributions among the three fractions for the added metals. The ratios of the M1 fraction to the M2 fraction are also calculated and presented in this table. For Cd(I1) and Zn(II), the M1 to M2 ratio is relatively large for the spiked sample and cannot be calculated in the original samples because no M2 fractions of these metals were detected. However, the M1 fraction for Zn(I1) is clearly dominant in the original sample from the data in Table 111. For Cu(I1) in the ditch water sample, the MI to M2 ratio for the spiked sample is reasonably close to that
ANALYTICAL CHEMISTRY, VOL. 61, NO. 6, MARCH 15, 1989
Table IV. Distribution of Cu, Cd, and Zn in Water Samples
sample Cu(I1) river ditch
Cd(I1) river ditch Zn(I1) river ditch
% Mla % M2
29 4 71 66 85 80
24 73 4 4 ND 5
% M3
[M1]/[M2lb
47 23 25 30 15 15
0.86 (0.17) 0.075 (0.11) 18 (-) 19 (-)
- (-)
20 (-)
[M31,' pg/L 3.54 (0.72) 2.23 (0.85) 1.53 (0.04) 1.84 (0.06) 1.37 (0.47) 1.09 (0.20)
a % M1, '70 M2, and % M3 are the percentages of M1, M2, and M3 species calculated for the added metals, respectively. [M1]/ [M2] is the ratio of the M1 concentration to the M2 concentration given in Table I11 for the spiked and original samples (in parentheses). e [M3] is the estimated M3 concentration for the spiked and original samples (in parentheses),using the relationship [M3] = (([Ml] + [M2])/(% M1 + % M2)) - [MI] + [M2].
of the original sample such that a reasonable estimate of the M3 fraction in the original sample can be made. In contrast, for Cu(I1) in the river water sample, the M1 and M 2 ratio of the spiked sample is about a factor of 5 larger than that of the original sample and the standard addition procedure cannot be used to estimate the original metal speciation in the water sample. In this case, the data suggest that there is a low concentration of natural complexing agents that form strong nonlabile complexes with Cu(I1) (M2 species). The metal spike might have exceeded the complexing capacity of this fraction of the complexing agents in the sample. A smaller Cu spike might have given a better estimate of the speciation in the original sample. The concentrations of the M3 species in the spiked samples and original samples are also calculated and given in Table IV for each metal under study. About 15-20% of the added Zn(I1) and 25-3070 of the added Cd(I1) in water samples appear in the M 3 fraction. Also 23% of the added Cu(I1) in the ditch water and 47% of the added Cu(I1) in river water appear in the M3 fraction. The M3 metal species include metals strongly associated with very large colloidal matter that do not dissociate in the Chelex-100 column and are not retained by the AG Mp-1 column (the molecular exclusion limit about 75 000 molecular weight). Guy and Chakrabarti (20) reported that about 30% of Cu(I1) in humic acid solution and river water samples were bound to species with sizes greater than 5.1 nm (or above 100000 molecular weight). Giesy and Briese (21) found that about 30% of Cd(I1) and Zn(I1) are associated with large size organic matter (300 000 molecular weight).
CONCLUSIONS The automated two-column ion exchange system presented here offers several advantages. First, the system has the incorporated preconcentration ability for two fractions of dissolved metal species. The concentration factor with a 10-mL sample loop is about 50 compared with direct flame atomic absorption (AA) measurement. Therefore, with a
529
10-mL sample loop, the present system is able to determine many trace metal species at the 0.1 rg/L level with the on-line flame AA spectrophotometer. Second, the present scheme for determination of metal speciation is simple and rapid. The automated on-line operation requires minimal sample manipulation compared with other reported schemes. The sample throughput of the two-column system is aboout 10 min per sample with a 10-mL sample loop. Third, two major fractions of dissolved trace metal species in natural waters are determined directly with a single aliquot of the sample solution in one sample run. In other measurement schemes, at least two separate measurements with two sample aliquots are usually required and one fraction is determined by difference. Fourth, the system provides a versatile tool to study the stability constants and kinetics of trace metal complexes in natural waters. One limitation of the method is that the third metal fraction cannot be directly determined. Although more studies are required, it appears that standard addition can be used to estimate the distribution of trace metals among the three metal fractions. Alternatively the third fraction can be determined by the difference between the total metal concentration in the water sample and the concentrations of the first two fractions. Registry No. HzO, 7732-18-5; Cu, 7440-50-8; Cd, 7440-43-9; Zn, 7440-66-6; Chelex-100, 11139-85-8; AG MP-1, 39319-99-8; EDTA, 60-00-4; NTA, 139-13-9; ammonium acetate, 631-61-8; glycine, 56-40-6; iminodiacetic acid, 142-73-4.
LITERATURE CITED Batley, G. E.: Florence, T. M. Anal. Lett. 1976, 9 , 329-338. Batley, G. E.; Florence, T. M. Mar. Chem. 1976, 4 , 347-363. Florence, T. M. Water Res. 1977, 7 7 , 681-687. Figura, P.; McDuffie, B. Anal. Chem. 1980, 5 2 , 1433-1439. Hart, B. T.; Davies, S. H. R. Estuarine Coastal M a r . Sci. 1981, 72, 353-374.
Laxen, D. P. H.; Harrison, R. M. Sci. TotalEnviron. 1981, 79, 59-82. Florence, T. M. Anal. Chlm. Acta 1982, 147, 73-94. Batley, G. E. I n Trace Element Speciation in Surface Waters and Its Ecological Implicatiofls; Leppard, G. G., Ed.; Plenum Press: New .. York,-1983: pp 17-36. Florence, T. M. Talanta 1982, 2 9 , 345-364. Liu, Y.; Ingle, J. D., Jr. Anal. Chem., in press. Marino, D. F.; Ingle, J. D., Jr. Anal. Chem. 1881, 53, 292-294. Product Catalog, Bio-Rad Laboratories: Richmond, CA, 1987. Filby, R. H.; Shah, K. R.; Funk, W. H. In Preceedlngs of 2nd International Conferences on Nuclear Methods In Environmental Research ;
Vogt, J. R., Meyer, W., Ed.; NTIS: Springfleid, VA, 1974; pp 10-23. Shuman, M. S.; Dempsey, J. H. J.-Water Pollut. Control Fed. 1977, 49, 2000-2006.
Montgomery, J. R.; Santiago, R. J. Estuarine Coastal Mar. Scl. 1978, 6 , 111-116.
Hiraide, M.; Tillekeratne, S. P.; Otsuka, K.; Mizuike, A. Anal. Chim. Acta 1985, 772, 215-221. Martell, A. E.; Smlth, R. M. Critical Stability Constants; Plenum Press: New York, 1974 Vol. 1. Stolzberg, R. J.; Rosin, D. Anal. Chem. 1877, 49, 226-230. Figura, P.; McDuftle, 8. Anal. Chem. 1979, 51, 120-125. Guy, R. D.; Chakrabarti, C. L. Can. J . Chem. 1976, 5 4 , 2600-2611. Giesy, J. P., Jr.; Briese, L. A. Chem. Geol. 1977, 2 0 , 109-120.
RECEIVED for review April 18, 1988. Resubmitted July 11, 1988. Accepted December 15, 1988.
