727
Anal. Chem. 1986, 58,727-731
Table IV. Metal Capacities' (mM/g) for Poly(dithi0carbamate) Resins with Different SCN Content resin sample
m g S/g (SCN)
1 2 3 4 5
2.07 1.70
0.98 0.53 0.16
CU
co
Fe(II1)
0.38 0.41 0.68 0.77 0.85
0.009 0.014 0.069 0.073 0.088
none none 0.001 0.002 0.0043
M e a n of triplicates. Relative standard deviations are 1.5-2.5%.
tation was observed when Cu(I1) solution was added to the resin. Because in the capacity measurements the resin is filtered then ashed, the copper capacity refers to copper attached to the dithiocarbamate groups on the resin plus copper reacted with the thiocyanate groups on the resin forming the Cu(1) thiocyanate precipitate. The precipitate is filtered with the resin, but since the precipitate is somewhat soluble, the results can be ambiguous. The capacity measurements are summarized in Table IV. The comparison of the Cu, Co, and Fe(II1) resin capacity demonstrates that the higher the thiocyanate content of the resin the lower the capacity for the metal. Presence of thiocyanate groups on the resin is very unfavorable for the binding of the above elements. All metals that form thiocyanato
complexes, e.g., most of the heavy metals, probably can react with the thiocyanate groups on the resin. Therefore, the metal uptake of the resin can be influenced by this reaction. In conclusion, the poly(dithi0carbamate) resin should be prepared in the presence of pyridine instead of ammonium hydroxide so that only dithiocarbamate functional groups are formed on the polymer. Registry No. SCN-, 302-04-5; Cu, 7440-50-8; Co, 7440-48-4; Fe, 7439-89-6.
LITERATURE CITED (1) Barnes, R. M. J . Test. Eva/. 1984, 12, 194. (2) Barnes, R. M. B o / . Trace Hem. Res. 1984, 6. 93. (3) Hackett, D. S.;Siggia, S. "Environmental Analysis": Ewing, G. W., Ed.; Academic Press: New York, 1977; p 253. (4) Hackett, D. S.Diss. Abstr. I n t . B 1977, 3 7 , 4430. (5) Burger, K. "Coordination Chemistry Experimental Methods"; CRC Press; Boca Raton, FL, 1973; p 98. (6) HorvBth, 2s.;Barnes, R. M.; Murty, P.S.Anal. Chim. Acta 1985, 173, 305. (7) Schulek, E. I n "Volumetric Analysis": Interscience: New York, 1957; p 303. (8) Murthy, R. S.S.; Horvlth, 2s.;Barnes, R. M., submitted for publication in J . Anal. Atm. Spectrosc. (9) "Gmelin Handbuch Der Anorganischen Chemie, Kohlenstoff Tell D4"; Springer-Verlag: Berlin, Heidelberg, New York, 1977; p 187.
RECEIVED for review September 25,1985. Accepted November 14,1985. This research was supported by the ICP Information Newsletter.
Solution Phase Complexing of Atrazine by Fulvic Acid: A Batch Ultrafiltration Technique Donald S. Gamble* Chemistry and Biology Research Institute, Agriculture Canada, Ottawa, Ontario K I A OC6, Canada, and Department of Chemistry, Concordia University, 1455 Boulevard de Maisonneuve Ouest, Montreal, Quebec H3G IM8, Canada
Mohammed I. Haniff and Raymond H. Zienius Department of Chemistry, Concordia University, 1455 Boulevard de Maisonneuve Ouest, Montreal, Quebec H3G l M 8 , Canada
A batch ultrafiltration method for measurlng atrazine complexlng In fulvic acid soluti6ns has been Investigated experlmentally. The method produces a dilution curve by uslng a set of dilutions of an atrazine stock solution. I t has been demonstrated experlmentally that thls batch dilutlon curve method may be used for three purposes. The first purpose was to make direct dlagnostic checks for ultrafiltratlonmembrane Interferences and other experimental errors. I f any ultrafiltration membrane interferences exist, they were undetected above the level of the random analytlcal chemical errors In the cases tested. The other two purposes that have been met were to investlgate the theory of methods for determining complexlng capacity and complexing equlllbrlum and then to demonstrate them wlth experimental data.
The phenomena to be examined during research on reversible labile complexing of organic pesticides in fulvic acid *To whom correspondence should be addressed a t Chemistry and Biology Research Institute, Agriculture Canada.
solutions may include the complexing capacity of the dissolved fulvic acid and equilibria at loading below site saturation. It is convenient to have experimental methods that yield information about both phenomema together. In such solution phase investigations, the first and most general analytical chemical difficulty is the problem of measuring separately the free and complexed forms of at least one reactant. For some types of samples, spectrometric and electrochemical methods are not available. This is especially true of fulvic acid solutions, in which organic pesticides or metal ions are complexed. At least a few authors have reported the use of ultrafiltration methods for investigations of metal ion interactions with humic materials (1-3). Buffle and Staub (4) and Staub et al. (5)have published an experimental study of the batch dilution curve method for metal ions, together with a theoretical analysis of partial membrane permeability to fulvic acid. Almost concurrently, Haniff et al. have applied the batch dilution curve method to an investigation of the solution phase complexing of atrazine by fulvic acid (6). The work of several authors has made it clear that the risk of membrane interferences must be heeded (7-10). These requirements suggest the objectives of this work. One objective is to devise and
0003-2700/86/0358-0727$01.50/0 0 1986 American Chemical Society
728
ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986
curve in Figure 1,obtained without fulvic acid, is an example of such a membrane test curve. For the theoretically ideal test curve, with no atrazine retained or rejected by the membrane
ce = CTo
(4)
Bo = 0
(5)
I*M/L
Bi = CJVo
(6)
It may therefore be exactly predicted according to eq 7, which c e
I
b
Ib
Ib
v;zb
2:
30
45
4b
b,
50
V,XIO~.ML
Figure 1. Batch dilution curves of an atrazine standard stock solution: (M) calibration curve and membrane test, with no fulvic acid, C, = 0.552 4- 1.45V,, f = 0.998; (e) experlmental curve with fulvic acld, complexing region, C, = 3.32 0.395 V,, f = 0.995; (A)experimental curve with fulvic acid, postcomplexing region, C = -16.2 -t 1.47V, f = 0.994, 25 f 1 OC. The standard stock solution concentration was C, = 1.1 X lo4 M. The aliquot volume of the standard stock solution is V , L ( v , x io3 mL).
+
demonstrate tests for the properties of ultrafiltration membranes. Another objective is to demonstrate the calculation of complexing capacities and complexing equilibria from the experimental parameters of the batch dilution curve method.
THEORY The dilution curve method employs a sequence of test or sample solutions having the same volume, into which have been put increasing aliquots of a standard stock solution of the reagent under investigation. Ultrafiltration and chemical analysis of each solution in turn then produce a dilution curve. In the absence of complexing agents and membrane effects, this is simply a series of dilutions of the standard stock solution. With the presence of a complexing agent, a type of titration curve results. Types of Ultrafiltration Dilution Curves. Typical examples of the two types of dilution curves are illustrated by Figure 1. There are membrane test curves, which exist without fulvic acid, and the experimental curves for which fulvic acid is present. An experimental curve contains two regions that are separated by a break point. The first region to appear as the experiment preceeds is the complexing region of the curve. After the break point comes the postcomplexing region, which has a curve apparently parallel to the corresponding membrane test curve. It is postulated that the slopes and intercepts of the three curve regions may be exploited for practical purposes. The nature and significance of each of them must therefore be examined. Membrane Test Curves. For V, liters of stock solution having a concentration, in FM, of C,, the initial mass of atrazine is woo = csv, (1) Dilution according to eq 2 then gives the initial concentration cTo = wT0/Vo = MOO MOB0 (2) in the ultrafiltration cell. Vois the initial sample volume in each of the ultrafiltration cells. Moo and M O ~ O are the molarities of free and bound atrazine, initially before ultrafiltration. The atrazine concentration measured in the filtrate is C,. When it is plotted against aliquot volume, then the experimental eq 3 is anticipated. B, and Bo are then the slope Ce = Bo + BIV, (3) and intercept calculated by linear least squares fit. The upper
+
=
CTo
=
(cs/vo)vs
(7)
is a simple rearrangement of eq 1 and 2. The experimental constants Bo and B1may therefore be used as diagnostic tests for membrane effects and analyical chemical errors. If the membrane either rejects or sorbs atrazine initially, with a subsequent breakthrough of atrazine into the filtrate, then the intercept Bo will have a negative value instead of being zero. If previously sorbed atrazine were to be desorbed into the filtrate, then a positive nonzero value of Bo would result. The slope B1could also be distorted by membrane effects. For example, either total rejection or total sorption would give
B1 = 0
(8)
Partial rejection or partial sorption would give the intermediate result
0 < B1
< C,/Vo
(9)
Desorption into the filtrate of previously sorbed atrazine would make the slope too large. That is
B1
'c,/vo
(10)
If there are no bias errors caused by membrane effects, then measured values of the slope and intercept will provide estimates of random experimental errors, independently of those that may arise during experiments with fulvic acid. Experimental Curves, Complexing Region. While nonlinear complexing curves may be possible, only linear cases were found in this work. Atrazine concentration measurements on the filtrate, Ce (pM), generally yield eq 11 for the Ce = Ba0 + BalVs
(11)
complexing region of an experimental dilution curve with fulvic acid present. Figure 1 shows an example in the first part of the lower curve, before the sharp break. B,, and BaO are the experimental slope and intercept. The theoretical value of the intercept is
BaO = 0
(12)
If the experimental values of BaOare not zero, then the diagnostic use of membrane test curves should identify the causes. The theoretical value of the slope depends on the chemistry of the sample. A case having total complexing of the atrazine, leaving none free, gives the zero slope of eq 13.
Bal = 0
(13)
If there is no complexing, then
B,1 = B1 = C,/Vo
(14)
A reversible complexing equilibrium reaction produces an intermediate result according to eq 15. This region of the 0
< B,, < B1
(15)
experimental curve may then be used for the calculation of
KB,the complexing equilibrium constant for the reaction of atrazine with fulvic acid.
ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986
729
Table I. Test Curves for YM2 Membranes Properties, No KCl PH
C, stock solution, pM
theor C, = Bo
B1, pM/mL of aliquot exptl
+ B1 Theoretical Values:
3.50 3.50 3.50 3.50
79.66 63.50 78.42 76.31
1.593 1.270 1.568 1.526
1.30 1.72 2.28 2.77 3.21 3.50 4.50 6.00 8.00
75.46 70.80 62.16 74.74 87.48 109.4 66.67 84.32 77.37
1.509 1.416 1.243 1.495 1.750 2.189 1.333 1.686 1.547
1.48 1.44 1.36 1.45 2.18 2.12 1.34 1.65 1.73
1.19 1.38 2.14 2.38 3.02 3.96 4.04 5.23
66.35 86.42 69.10 72.35 66.21 69.72 78.85 59.28
1.33 1.73 1.38 1.45 1.32 1.39 1.58 1.19
1.32 1.78 1.37 1.55 1.25 1.41 1.35 1.13
% error
Eo; pM exptl
type of experiment
E , = 0 pM, B1 = (C,/50.00) pM/mL
1.56 1.08 1.31 1.61
-3.3 -7.5 -8.2 5.5
0.435 3.36 6.42 -1.95
-2.19 2.40 9.41 -3.01 24.6 -3.15 0.53 -2.14 11.8
-1.43 -2.33 -7.37 0.552 -4.85 2.62 -0.282 3.72 -9.02 4.62 -1.45 3.01 -0.61 5.94 1.72 5.59 7.33
-0.75 2.9 -0.72 7.6 -5.3 1.4 -14.6 -5.0
variable fulvic acid concn
variable PH
functional group blocking by Cu(I1)
chelation
" B omean = 0.763 pM, u = f4.41.
Experimental Curves, Postcomplexing Region. This experimental region is generally characterized by straight lines having negative intercepts. These negative values correspond to the breakthrough aliquot volumes that occur when the available complexing sites have been just exactly used up. The breakthrough volumes therefore represent compleximetric titration end points. In Figure 1, the postcomplexing region is seen as the second part of the lower curve, after the sharp break. Bbl and BbOare the linear least squares fitted slope and intercept. ce
= BbO + BblVs
(16)
In eq 16, the theoretical values for the constants are BbO
Bbl
17.5 >31 >18 >12.5 >7.5 >12.5
0.978 1.36 1.47 2.19 2.08 1.30
-17.4 -20.4 -16.2 -19.4 -23.5 -3.3
-20.7 -19.3 -19.5 -19.6 -25.4 -2.35
0.1 M KC1
1.27 1.83 2.77
>33 >23 >26
0.966 1.47 1.75
-15.8 -16.9 -16.3
-19.2 -23.9 -16.4
Cu(I1)
1.19 1.38 2.14 2.38
>12 >24 >17 >14
1.06 1.88 1.44 1.58
0.826 -18.9 -4.99 -10.9
-5.17 -17.6 -9.34 -17.10
expts
yielded very low correlation coefficients. The conclusion drawn from about 3 dozen membrane test curves is that the experimental errors are mostly random analytical chemical
ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986
Table V. Complexing Capacities: Calculated from the Slopes and Corrected Intercepts of the Postcomplexing Region (See Table VI) V, end dissolved metal ions
pH
C, stock solution, p M
point,
mL
complexing capacity, mmol/g of fulvic acid
VE = -(BbO/Bbl) mL 31.9 17.7 19.8 19.5 26.8 2.41 -2.68 0.46
none
1.30 2.28 2.77 3.21 3.50 4.50 6.00 8.00
75.46 62.16 74.74 87.48 109.4 66.67 84.32 77.37
21.2 14.2 13.3 8.95 12.2 1.81 -1.59
0.1 M K C l
1.27 1.83 2.77
68.51 75.00 79.32
27.2 24.4 14.9
3-82
78.64
19.9 16.3 9.37 0.0
1.19 1.38 2.14 2.38
66.35 86.42 69.10 72.35
4.88 9.38 6.49 10.82
6.48 16.2 8.97 15.7
Cu(I1) expts
0.30
o.n
Table VI. Weighted AvTrage Equilibrium Function for Atrazine Complexing, K g : From Theoretical and Experimental Values of B ,
KB pH
theoretical KB =
exptl
( B , - B,1)IB,1C0
error f r o m B,, %
e x p t l conditions
co = 1.000 g / L
731
of the intercepts BaOpermit the empirical corrections to be made in the subsequent compleximetric end point calculations. The substantial sizes of the nonzero values of BaOobviously make the empirical calculations important. Table VI illustrates the manner in which the postcomplexing region of the experimental curves is corrected in preparation for the calculation of complexing capacities. The errors in the slope of Bbl are comparable to those in the slopes of the membrane test curves, and the origins may be the same. The intercepts Bw are corrected with the BaOdata from Tables I11 to V. To the extent that the two regions of the experimental curves do not have identical biases, the corrected Bm values will still retain residual errors. A substantial error reduction should none the less have been achieved. According to the calculation outlined in Tables V and VI, the total error in the calculated complexing capacity will include both this residual error and the analytical error in the stock solution concentration. Whenever the calculations must be based on gas chromatography measurements, the analytical errors of 1to 2% give errors in the complexing capacity of 10 to 2 0 % (13). HPLC or spectrophotometric methods may therefore prove to be preferable. An examination of eq 28 together with the magnitudes of the data to be used with it suggests that the error in B1 should contribute the largest portion of the error in RB.This is confirmed in the present case by a comparison of the errors listed in Tables I and I1 with those of Table VI. The most important opportunity for further development of the batch dilution curve method will come from the refinement of analytical chemical methods. High-pressure liquid chromatography and spectrophotometric methods should be examined for this purpose. Registry No. Atrazine, 1912-24-9.
1.30 1.72 2.28 2.77 3.21 3.50 4.50
5.26 1.11 0.724 2.79 1.69 6.82 0.149
5.14 1.14 0.886 2.67 2.34 6.57 0.155
-2.3 3.3 22.4 -4.1 38.9 -3.6 4.1
n o KC1 n o Cu(I1)
1.27 1.83 2.77
2.66 2.30 0.404
2.90 2.36 0.549
9.0 2.9 36.0
0.1
1.19 1.38 2.14 2.38
1.20 0.466 0.572 2.39
1.18 0.509 0.561 2.62
-1.4
Cu(I1)
M KCl
9.1 -2.0 9.8
errors, with no membrane interferences having been detected. Linear least squares fit results for three types of experiments in the complexing region are listed in Table 111. The slopes, Bal, can only be used in this case for determining whether complexing is total, partial, or absent. The nonzero values
LITERATURE CITED (1) Guy, R. D.; Chakrabarti, C. L. C a n . J. Chem. 1976, 54, 2600. (2) Hoffman, M. R.; Yost, E. C; Eisenreich, S. J.; Maier, W. J. Envlron. Sci. Techno/. 1981, 54, 655. (3) Tuschall, J. R., Jr.; Brezonik, P. L. I n “Aquatic and Terrestrial Humic Materials”; Christman, R. F., Gjessing, Eds.; Ann Arbor Science Publishers: Ann Arbor, MI, 1983; Chapter 13. (4) Buffle, J.; Staub, C. Anal. Chem. 1984, 56, 2837. (5) Staub, C.; Buffle, J.; Haerdi, W. Anal. Chem. 1984, 56, 2843. (6) Haniff, M. I.; Zienius, R. H.; Langford, C. H.; Gamble, D. S. Sci. Health, Part B 1985, 820, 215. (7) Blatt, W. F.; Robinson, S.M.; Blxler, H. J. J. Anal. Blochem. 1968, 26, 151. (8) Grice, R. E.; Hayes, M. B. H. Proc. Br. Weed Control Con!., 77th 1972, 784. (9) &Ice, R. E.; Hayes, M. B. H.; Lundie, P. R.;Cardew, M. H. Chem. Ind. (London) 1973, 233. (10) Smedley, R. J., personal communication, 1984. (11) Gamble, D. S. Can. J . Chem. 1972, 5 0 , 2680. (12) Schnitzer, M.; Skinner, S. 1. M. Soil Sci. 1963, 96,86. (13) Gamble, D. S.;Khan, S. U.; Tee, 0. S. Pestic. Sci. 1983, 74, 537.
RECEIVED for review July 30,1985. Accepted October 30,1985.
