Fluorometric Reaction Rate Method for Determination of Silver R. b. Wilson' and J. D. Ingle, Jr." Department of Chemistry, Oregon State Universiv, Corvallis, Oregon 9733 1
The development and optimlration of the fiuorometrlc reactlon rate method for determination of Ag' Is described in detall. The method Is based on the enhancement by Ag' of the reactlon between oxlne-5-sulfonic acid and persulfate. This technlque provides a linear dynamic range of 6 ppb to 30 ppm for the trace determination of Ag. The relative standard deviation for the Ag analysis Is typically less than 0.5%. A number of metals Interfere with the technique. For the analysis of Ag in real samples, a dlthlzone extraction procedure was developed to eliminate Interferences and thls procedure Is applied to NBS Zn spelter samples.
Many analytical techniques are based on the reaction of the analyte with an organic reagent to form an absorbing or fluorescence product. Usually such reactions are allowed to go to completion and the final absorbance or fluorescence signal is proportional to the initial analyte concentration. In some cases these reactions are relatively slow (5 min or more to reach equilibrium a t room temperature), require heating, the product is unstable, and there are interferences due to other species reacting with the organic reagent to form products with similar spectral properties as the product formed with the analyte. Under these conditions, kineticsbased measurements may be advantageous (1-8) in terms of speed, accuracy, and specificity. Use of molecular fluorescence to monitor reactions in kinetics-based measurements has the potential advantage of greater sensitivity compared to molecular absorbance monitoring if the fluorescence product has a reasonable quantum efficiency. Also more selectivity is achieved since only a small fraction of molecules significantly fluorescence, both excitation and emission wavelengths may be chosen, and differences in non-analyte absorption have a smaller effect (9). Differences in the scattering signal or background fluorescence signal among different samples limit the accuracy of trace fluorescence equilibrium-based measurements. As long as these signals are constant for a given sample, they do not affect kinetics-based measurements since only the change in fluorescence signal is measured. Because initial rate measurements are made in the first few percent of the reaction, the concentration of the fluorescence species is much smaller than a t the completion of the reaction and the importance of the numerous effects which cause a nonlinear relationship between the fluorescence signal and the fluorophore concentration (9) are significantly reduced. It is clear from the above discussion that kinetics-based analysis with fluorescence monitoring is potentially a specific and sensitive technique for trace analysis. This paper is concerned with the critical experimental evaluation of this potential. Specifically, a previously described fluorometric reaction rate instrument (10) is used for the kinetics-based determination of trace levels of Ag'. Ryan and Pal (11)determined Ag by an equilibrium-based molecular fluorescence technique involving a reaction between 'Present address, Department of Chemistry, University of Wisconsin-Eau Claire, Eau Claire, Wis. 54701. 1066
ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
Ag', 8-hydroxyquinoline-5-sulfonic acid (OXSA) and potassium peroxydisulfate (Sz02-). .The fluorescence intensity reached a maximum in about 1h, remained constant for an hour, then increased. The fluorescence intensity was directly proportional to [Ag'] from 12.5 ppb to 5 ppm when the measurement was made between 60-90 min after mixing and the OXSA to metal mole ratio was 101. Low results were obtained for ratios less than 101and ratios greater than 1OO:l were not tried. The fluorescence was measured at 485 nm after excitation at 375 nm. Interference studies for 0.4 ppm Ag' with 22 cations and 4 anions revealed that more than 1 ppm of Cu2+,Hg2+,and Pd2+quenched while E4' and HP+enhanced the fluorescence. Apparently Concentrations greater than 75 ppm A13+,50 ppm La3+, 20 ppm Co2+,and 10 ppm Ni2+and Fe3+ caused some problems which were not specified. EDTA, ethanol, acetone, and dioxane destroyed the fluorescence. The nature of the reaction was not known but two possibilities were postulated. The fluorescent product could be (1)a stable silver(III)4XSA chelate or (2) an oxidized product of OXSA. The latter possibility was favored. Since i t takes about a n hour t o reach equilibrium, it was felt that this system could possibly be adapted to fluorometric kinetics-based method of analysis. This system was chosen for further study to gain knowledge of the chemistry involved, to develop a fluorometric kinetics-based method of analysis for silver and to critically evaluate the fluorometric reaction rate instrument (10) and technique.
EXPERIMENTAL Solution Preparation. All glassware and Telfon bottles were cleaned as previously described (10). All solutions were prepared with double distilled water produced by a Corning AG-3 still connected to the house distilled water. All chemicals used were Analytical Reagent grade or better. All solutions were stored in black Teflon bottles. The following is a list of the stock solution preparation procedures: 1. 8-Hydroxyquinoline5-sulfonic acid (OXSA) 2. Potassium peroxydisulfate 3. Blank and Ag solvent
4. Silver(1) 5. EDTA
6. Dithizone (HDz)
100 ppm = 0.1000 g OXSA/l L H,O 0.1 M = 27.0320 g K,S,O,/ L H,O 0.018 N H,SO, = 1.00 rnL H,SO, (concn)/l L H,O 100 ppm = 0.1575 g AcNOJl L 0.ois N-H,SO, 5%= 5 g EDTA/100 mL H,O 0.01% = 0.0100 g HDz/100 mL CCl,
Lower concentrations of the above were made by serial dilution. Some problems were encountered which should be noted. First, the OXSA and Sz02- concentrations listed above are the practical upper limits due to solubility limitations. The 100 ppm OXSA solution required heating to speed dissolution. Without heating, dissolutiontook overnight. OXSA can be made more concentrated if dissolved in 0.1 N NaOH or KOH; however, this proved unacceptable for analysis since the reproducibility was poor, apparently due to neutralization problems. The SZOs2-solution required continuous shaking for several minutes for complete dissolution.
No significant differences were noted whether or not the OXSA, K2S208,or AgN03were dried in an oven at 110 "C for a few hours before weighing. Instrumentation and Measurement Conditions. The fluorometric instrument designed specifically for kinetics-based measurements has been described (10). All fixed wavelength measurements were made with the excitation monochromator set at 366 nm with 2-mm slits (17-nmspectral bandpaas) and a 480-nm interference filter (half-width of 17 nm). For final rate measurements, the ratemeter (12) was set for measurement and integration times of 32 and 16 s, respectively. Measurement times up to 5 min could be used if desired but do not yield any significant improvement over 32 s. A delay time of 64 s was chosen so that the induction period for the lower concentrations would be completed and the measurement would be made in the linear portion of the voltage vs. time curve for all concentrations. All measurements were made at 25 "C. The order of addition of the various reagents into the sample cell was: 1 mL OXSA, 1 mL Ag' or blank, and 1 mL of S20e2solution. This order of addition was varied and no significant difference was found for any combination. The procedure for introducing and mixing the reagent and sample solutions is critical and has been discussed (10). Optimization Studies. To optimize the reagent concentrations and obtain their orders, the initial concentration of one of the constituents was varied while the rest were held constant and the initial reaction rate was measured. A plot of the log of the initial rate vs. the log of the initial concentration of the species of interest yields a curve whose slope at a given concentration is equal to the order with respect to that species at that concentration. Ideally for kinetics-based measurements, the optimum concentration of each reacting constituent, except the analyte, will be the one which yields the smallest relative standard deviation for the rate measurement and is in a region which is zero order in concentration with respect to that species. The latter condition is desired so that small fluctuations in concentration will have no effect on the initial rate. Conditions should also be chosen so that the initial rate is first order with respect to the analyte. To obtain kinetic and optimization information and a calibration curve, the above procedure was repeated for each initial concentration at least five times to obtain a mean, standard deviation and relative standard deviation for the initial rates of the background and analyte plus background reactions. The background reaction is the reaction between OXSA and Sz02' in the absence of Ag'. Each kinetics study was repeated at least three times to ensure the reproducibility of the results. For each species the log-log plot of initial rate vs. initial concentration is made to obtain information about the orders of each constituent. The concentration yielding the lowest relative standard deviation for the initial rate of the background reaction and an acceptable order with respect to the reagent was taken as optimal. Interference Studies. Interference studies were carried out for 24 metal and 6 nonmetal ions. The effect of each ion at different concentrations was tested by substituting a solution of the ion for the Ag' solution and also a solution containing 0.5 ppm Ag' plus the potentially interfering ion. If the ion in question gave an interfering initial rate, the ion concentration was reduced by serial dilution and the procedures were repeated. This was repeated until no effect (no interfering initial rate) was observed. That is, the initial rate with and without the ion solution for the background and Ag plus background reactions were within one standard deviation. Extraction Procedure. Because a number of ions were found to interfere, a number of separation schemes were tried. The method of interference elimination that was finally utilized involves the well known (13-16) solvent extraction of Ag' with dithizone (HDz) or diphenylthiocarbazone in CCl,. Dithizone can be made selective for Ag' by adjusting the pH of the aqueous phase below zero and extracting in the presence of EDTA. The EDTA effectively complexes metals other than Ag so that they will not be extracted or at least extracted very slowly. The Ag is only very weakly complexed by EDTA so that the HDz effectively competes for the Ag which is extracted almost instantaneously. The extraction procedure is as follows:
1. One mL of a Ag' standard or sample solution containing Ag' was diluted with 20 mL of 5 N &SO4 to adjust the pH below zero, thus increasing the Ag' selectivity. 2. Ten mL of 5% EDTA was added to the Ag' solution to mask those ions (e.g., Hg, Au, Pd, Cu) that can still be slowly extracted at very low pHs. As mentioned above, Ag is easily extracted in the presence of EDTA. 3. The Ag' was extracted twice with 5 mL of 0.001% HDz in CCl, for about 15 s each. The short extraction time also enhances the selectivity for Ag since it is rapidly complexed by HDz, whereas under the extraction conditions other metals (above) are only slowly extracted. 4. The CC14 layers were separated from the aqueous layer, added together and evaporated down to about 0.5 mL. Evaporation to dryness is avoided because this causes large losses of Ag' (17). This is because Ag' at elevated temperatures in the presence of charred organic material is reduced to elemental Ag. 5. To this 0.5 mL of silver-dithizonate in CC14,about 2 mL of 30% HzOzwere added and then evaporated to dryness. This step is necessary to destroy the HDz to prevent interference in the Ag-OXSA-S20s2-reaction since it is easily oxidized by the Sz02-. In contrast to step 4, this evaporation does not cause appreciable loss of Ag since the organics were destroyed by the HzOz. 6. The residue from step 5 is then dissolved in 0.018 N HzSO4 and brought up to the mark in a 10-mL volumetric flask. No pH adjustment is required after this step. The final concentration of that in the original sample. is 7. The silver content is then measured as described in a previous section. Zinc Spelter Analysis. To verify the viability of the analysis technique including the extraction procedure, an analysis was made of the Ag content of the National Bureau of Standards, Zn Spelter standard number 109. The procedure was as follows: 1. A 0.5-to 1.0-g NBS No. 109 standard is weighed out. 2. The sample is dissolved in 5 mL of 8 N HN03 and then evaporated to about 1 mL. The presence of as little as 2 N HN03 will oxidize the HDz making it useless for silver extractions. Hence, it is important that most of the HNOBbe fumed off. 3. The extraction and measurement procedures were then performed as previously described.
R E S U L T S A N D DISCUSSION Equilibrium Studies. An equilibrium study was made of the Ag-OXSA-S202- system in order to make a comparison with the work previously described (11)and to help determine the proper conditions under which the initial rate measurements should be made. It was discovered that a background reaction occurs between OXSA and Sz02- in H2S04. This is believed to be due to the formation of an oxidized product of OXSA by the Sz02-,The previous researchers (11) did not mention the presence of a background reaction. Excitation and emission spectra were also taken in order to obtain the maxima for excitation and emission, 366 and 485 nm, respectively. An excitation wavelength of 366 nm was optimal in our system because a Xe-Hg lamp was the source. The uncorrected emission spectra of the fluorescence species produced in the Ag-OXSA-S202- system is shown in Figure 1. The excitation and emission spectra of the species formed in the background reaction with OXSA-H2S04-Sz02-, were identical in shape and position to those with Ag' present. A comparison of the emission spectrum in Figure 1 to that previously reported (11)indicates that they are the same (e.g., halfwidth = 90 nm and = 485 nm). The length of time the reaction takes to reach the maximum fluorescence signal (Ef"")and the length of time it remains stable (fluorescence signal constant within 1%) before decreasing, are dependent on both the Ag and OXSA concentrations as shown in Table I. The p H and persulfate concentration were held constant for all the equilibrium studies. All concentrations specified in this work are initial concentrations. Cell concentrations are 1 / 3 the initial since 1 mL of each reagent is used. ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
1067
I
'
I
I O
->
8
E ' 6 r
w
4
2
350
400
450
500
A,,,,
(n m)
550
600
Figure 1. Ag'-OXSA-S2082- emission spectrum. [Ag'] = 1 ppm [S20s2-] = 0.1 M, [OXSA] = 10 ppm, R, = lo7 0,T = 0.1 s, ,EpMT = 800 V, A,, = 366 nm, scan rate = 100 nm/min, excitation slit = 2.0 nm, emission slit = 0.5 nm
Table 11. Summary of Ag-OXSA-S,O,"- Kinetics Data Background reaction: v o a [S,O,l-]I for 1.6 X M < [ S 2 0 8 2 -e] e 9.4 x 1 O + M u o a [OXSA]"' for [OXSA] e 3 ppm v o a [OXSA]' for 3 ppm e [OXSA] e 30 ppm v o a [H']' for pH 2.2 Ag+ enhanced reaction less background reaction: v o a [A&]' for 6 ppb e [Ag'] e 30 ppm u o a [S,O,~-]lfor [S,O,=] e 3.1 x M u o a [S20,2-]'/2for [S,O,=] 'r 3.1 X lo-' M v o a [OXSAIo for 1 pprn [OXSA] e 1 0 ppm u o a [OXSAI-I'' for [OXSA] B 10 ppm v o a [H']' for 2.8 9 pH > 1.7 Rate laws under analysis conditions: [S,O,l-] = 0.1 M, [OXSA] = 10 ppm, pH = 2.2 (1) Background reaction: u o = kobs[Sz082-] (2) Ag+ enhanced reaction less background reaction: V o = hobs [ A g 1 [s20,2-] (3)
Table I. Summary of Ag-OXSA-S,0,2Equilibrium Dataa [OXSA], [Ag'], PPm PPm
Efmm time, min
Stable time, min
4.0
-
3.0
-
2.0
-
1.0
-
v)
EP-,
mV
3
0
30 30 10 10 10 1 1
10 0
10 0.3 0
10 0
127 145 34 110 131 3.5 16
8 10 3 5 12
1 0.25
3280 3040 1350 1030 910 117 96
a Final p H = 2.2, [ S 2 0 8 2 - ] =0.1 M, R f = l o 6 a , E,, 800 V without ratioing.
0
W
5
= 0.0
As can be seen from the table, the time it takes to reach Ef"" increases with increasing OXSA concentration with or without Ag', is shorter when Ag is present, and is also shorter a t higher Ag' concentrations. The stability time also follows a clear trend, increasing with increasing OXSA concentration in the presence of Ag' and at constant OXSA concentration decreasing with increasing Ag' concentration. Previous data (11)indicate that the fluorescence signal increases after the stable period and that the stable period is an hour which both conflict with our data. From these data it is clear that the use of this system for an equilibrium-based technique for Ag is suspect. During the course of the equilibrium studies emission spectra of both the Ag and the background reactions were taken at times before, during, and after the maximum fluorescence signal was reached. In all cases, the fluorescence spectra for the background reactions were identical to those for the Ag reactions. These data support the idea of an oxidation reaction rather than a complexation reaction taking place since the fluorescence product appears to be the same with or without Ag' since without Ag' or other metals, no complex can be formed. The equilibrium data cannot directly indicate the nature of the reaction, but clearly Ag' enhances the rate of the reaction and shortens the time to reach equilibrium. Preliminary data suggest that the background reaction depends on the thermal decomposition of Sz0;- and that Ag' acts as a catalyst in accordance with other studies involving S z 0 2 - and Ag' (18-26). Kinetics and Optimization. The kinetics and optimization study of the Ag-OXSA-S20s2- system was made as 1068
ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
0.0
1.0 LOG [s20i]
2 .o ( M x
3.0
4
io4
Figure 2. Log-log plot of [S20z-] vs. initial rate. a = Ag reaction less the background reaction, b = background reaction, [Ag'] = 10 ppm, [OXSA] = 6 ppm, pH = 2.5, R, = 10' 0,7 = 1 s, A,, = 480 nm (interference filter), other conditions are the same as for Figure 1
described in the Experimental section. The experimental rate information is summarized in Table 11. The data for the dependence of the initial rate and precision on the persulfate concentration are shown in Figure 2. The optimum SzOzconcentration (for the Ag' reaction less the background reaction) was found to be 0.094 M. At this concentration, SzOthas an order n equal to 'Izand the lowest relative standard deviation (RSD) is achieved. Although n = ' 1 2 is not ideal, it is better to have the initial rate dependent on the square root of the Sz02- than be linearly dependent. The range of RSD over the concentration range studied for the background reaction was 1.2-42% and for the Ag reaction, 0.4-7.2%. The precision continually decreased as the [SzO;-] was lowered. The results for the OXSA study are shown in Figure 3. The optimum OXSA concentration was found to be 10 ppm because it provided the lowest RSD and was zero order for both the Ag and background reactions. The range of RSD over the concentration range studied for the background reaction was 1.2-0.37% and for the Ag reaction, 0.92-0.23%. The results for the [H'] study are shown in Figure 4. The final pH was adjusted by making dilute HzS04 solutions (which were used as the blank and Ag' solvent) such that the final [ H'] (after mixing with Sz02- and OXSA) are as
5.0 1
I
I ,
2.0 0.0
1.0
2 .o
L O G [OXSA]
3.O
(pprnx
4.0
IO)
Flgure 3. Log-log plot of [OXSA] vs. initial rate. a = Ag reaction less the background reaction, b = background reaction. [Ag'] = 10 ppm, [S,O,'-] = 0.1 M, pH = 2.5, other conditions are the same as in Figure 2
1
0.0 I 0.0 5.0
I.o
J 2.0
,
3.0
4.0
5 .O
Figure 5. Ag calibration curve. [OXSA] = 10 ppm, [S202-] = 0.1 M, pH = 2.2, other conditions are the same as in Figure 2
v)
Table 111. Summary of Ag' Interference Study Data
4.0 3 0 0 (3
o
Species 3.0
-I
2.0 I.o L 0 G]'H[
2 .o
3.O
( M x l O
4.0
5
Flgure 4. Log-log plot of [H'] vs. initial rate. a = Ag reaction less the background reaction, b = background reaction. [Ag'] = 10 ppm, [OXSA] = 10 ppm, [S20,'-] = 0.1 M,other conditions are the same as in Figure 2 specified. A final pH of 2.2 (corresponding to an initial H2S04 concentration of 0.018 N) was optimal since it resulted in the best precision and a zero-order dependence on H' for both the Ag and background reactions. The range of RSD was 1.6-0.21% for the background reaction and 0.57-0.16% for Ag. A study was also made to determine the dependence of the reaction on the type of acid used. No difference was found between H2S04,H3P04,and HClO,; however, HNOBinhibited the reaction. Various concentrations of the first three acids were tried and the results were in agreement with those in Figure 4. The Ag concentration study yielded two results, the order of the initial rate with respect to the Ag' concentration and the Ag' calibration curve. Results are given in Figure 5. The initial rates plotted on the Ag' calibration curve are the difference between the Ag enhanced rate and the background rate. Relative precision ranged from 0.44-0.24% which indicates the overall excellent precision characteristics of this procedure and instrumentation. The detection limit, defined as the concentration yielding a rate the 2/2 times the standard deviation of the background reaction rate, for Ag' is 6 ppb. S/N calculations and ratemeter testing procedures (IO)indicate that the precision is limited by signal shot noise in the fluorescence signal below 1 ppm Ag' and by sampling and mixing reproducibility above 1 ppm Ag'. Twenty analyses may be run per hour. Interference Study. The interference study on the Ag-OXSA-S202' system was made as described earlier. A
WI) K(I) Mg(I1) Ba(11) Sr(11) Pb( 11) Ca(I1) Cu(I1) Fe(111) Hg(II) Cr(II1) c o (11) Ni(I1) Zn(I1) Cd(11) Mn(I1) Ti(I1) V(IV) Sn(11) Sn(1V) Al(II1) Zr( IV) Hf(1V) Pt(I1) Au( 111)
so,,-
PO,,c10,-
NO3c1F-
Concn, ppma 300 300 300 300
300 300 30 30 3 3 30 3 3 3 3 3 3 0.03 0.3 0.3 0.3
Concn, ppmb
...
...
...
...
0.1 0.3 0.03
150(1,(0.95)) 300(1,(0.13)) 30(1,(0.70)) 30(1,(0.52)) 1 5O(I, (0.98,O. 88)) 30(E2(1.1)) 30(E2(1.4)) W E , ( 1 . 1)) 30(1,(0.95,0.77)) 30(E3(1.01,1.08)) 30(I,( 0.98,0.92)) 0.3(1,(0.96,0.80)) 3(E,) 3(E,) 3(E2(1.1 )) 0.1(12(0.95)) W2(0.76)) 3(1,(0.93,0.70)) 0.3(E,(1 . 2 6 , l . l ) )
0.18 M 0.15 M 0.12 M 10-3 M 10-4 M 0.01 M
0.16 M( 1,(0.95,0.42)) 10' M( I, (0.9 5,O.88)) 0.5 M(E,(1.2))
0.01
... ...
a Concentration which had no effect on the rate of the blank or Ag enhanced reaction. Concentration which increased or decreased the rate of either the blank or Ag enhanced reaction. E, = Ion enhances the Ag reaction more than the background reaction. E, = Ion enhances both the Ag and background reactions equally. E, = Ion enhances the background reaction more than the Ag reaction. I, = Ion inhibits the Ag reaction more than the background reaction. I, = Ion inhibits both the Ag and background reactions equally. Relative increase or decrease of blank and Ag enhanced (0.5 ppm Ag+)rates are given in parentheses,
summary of the results is given in Table 111. It was hoped that many of the interfering ions which react would do so at a much slower or faster rate than with Ag so they could be differentiated against. Although the rates of reaction did differ ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
1069
Table IV. Summary of the Zinc Spelter Analysis Data“ RSD Ag found, Mode
ppmb
Fluorescence reaction rate Atomic absorption
0.772 0.735
each
sample, %c 0.38
..
*
RSD six Re1 Sam% ples, % error 16 21
3.5 8.1
After correcting for 94% extraction a Six sample. At least 5 runs for each sample. efficiency. somewhat, from Table I11 there are obviously many interferences. It should be noted that many of these ions are interferences only for determining Ag’ near the detection limit and not for higher Ag’ concentrations (e.g., 1 ppm). Also many of these ions interfere only at relatively high concentrations (greater than 1 ppm) which would not be found in many real samples. The most troublesome interferences are V(IV), Sn(II), Sn(IV), Al(III), Zr(IV), Hf(IV), Pt(II), Au(III), NO3-, and C1-. Dithizone Extraction. The procedure for the dithizone extraction of Ag described earlier was applied to Ag’ solutions in the range of 0.3 to 10 ppm. Measurements of the concentration of the standard Ag’ solutions before and after extraction were made with the fluorescence reaction rate instrument and a Varian Techtron AA-6 Spectrophotometer. The mean and relative standard deviation of the extraction efficiency at different Ag’ concentrations was 94% and 3%, respectively. Since extraction of the blank solution gave the same results as a nonextracted blank and since the extraction efficiency for Ag’ was reproducible over the concentration range tried, it was not necessary to prepare a separate Ag calibration curve for extracted Ag’ standards. Zinc Spelter Analysis. The NBS No. 109 standard Zn sample certification sheet indicates that it also contains 20 ppm Pb, 18 ppm Cd, 6 ppm Fe, 5 ppm Cu, 2 ppm Sn, and 0.8 ppm Ag. Hence the Zn spelter sample contains many of the metal ions which were identified to interfere with the reaction rate technique for Ag. The sample consisted of small metal filings. There was some difference in the coloration of some of the filings so that to obtain a homogeneous sample, the filings were thoroughly mixed, spread out on a piece of paper, and randomly transferred to weighing paper. The results of the analysis are shown in Table IV for the fluorescence reaction rate method and by AA analysis (Varian AA-6). As can be seen, the fluorescence reaction rate technique is very precise (0.38% RSD) for measurement of the Ag content in a given sample solution and the extraction procedure minimizes interferences so that an accurate measure of the Ag’ concentration is obtained. Since the RSD for the difference in the means of all six samples is 42 times larger than the average for each individual sample, it is obvious that the sample preparation procedure is the limiting factor in the analysis precision. Quite possibly the sample inhomogeneity is the largest contributor to the RSD since the RSD for repetitive extractions on a given Ag standard (3 ppm) is only about 3 % . CONCLUSIONS The kinetics of the Ag-OXSA-S202- system was studied in detail and a new analytical method for trace Ag determinations was developed. Conditions for Ag analysis were chosen for highest precision and for smallest reagent con-
1070
ANALYTICAL CHEMISTRY, VOL. 49, NO. 7, JUNE 1977
centration dependence. Experimental rate laws were determined to help understand the reaction in question. The exact nature of this reaction is not known but the evidence suggests that the product is an oxidized form of OXSA and that Ag’ acts as a catalyst. Possibly the product is a quinone type compound since in alkaline (pH 10-14) solution phenols are known to be oxidized by S20:- to quinones (27). The initial rate of the Ag-OXSA-S202- reaction is first order in [Ag’] over a large dynamic range, 6 ppb to 30 ppm, and depends on the square root of the S2O2-concentration. Because many metal ions interfere with Ag’ determinations, a dithizone extraction procedure was developed to make the technique specific for Ag’. Molecular absorption of the dithizone complex of Ag, dc arc emission spectroscopy, and AA are the most common techniques used for silver determinations (28). Compared to the proposed technique, the detection limit is considerably higher with molecular absorption and about the same with emission. Compared to AA, the detection limit is about the same, the dynamic range is much larger, and the total analysis would be longer for samples which require the extraction procedure. Typically, the relative standard deviation for rate measurements is about 0.4%. S/N calculations reveal that the rate measurements are limited by signal shot noise and sampling and mixing reproducibility (IO). LITERATURE CITED (1) H. V. Maimstadt, C. J. Dehney, and E. A. Cordos, C f i . Rev. Anal. Chem., 2, 559 (1972). (2) H. A. Mottola, Crif. Rev. Anal. Chem., 5, 229 (1975). (3) H. B. Mark, Jr., and G. A. Rechnitz, “Kinetics in Analytical Chemistry”, Why-Interscience, New York, 1968. (4) K. B. Yatsimerskii, “Kinetic Methods of Analysis”, Pergamon Press, Oxford, 1966. (5) H. 6. Mark, Jr., L. J. Papa, and C. N. Reiiley, Adv. Anal. Chem. Insfrum., 2, 255-385 (1963). (6) W. J. Biadel and G. P. Hicks, Adv. Anal. Chem. Instrum., 3, 126-140 (1964). (7) H. L. Pardue, Adv. Anal. Chem. Instrum., 7, 141-207 (1968). (8) S. R. Crouch, Comput. Chem. Instrum., 3, 107-207 (1973). (9) J. D. Winefordner, S. G. Schulman, and T. C. O’Haver, “Luminescence Spectrometry In Analytical chemistry”, John Wlley, New York, 1972. (10) R. L. Wilson and J. D. Ingle, Jr., Anal. Chem., 49, preceding paper. (11) D. E. Ryan and B. K. Pal, Anal. Chim. Acta, 44, 385 (1969). (12) R. L. Wilson and J. D. Ingle, Jr., Anal. Chlm. Acta, 63,203 (1976). (13) J. Korkisch, “Modern Methods for Separation of Rare Metal Ions”, Pergamon Press, New York, 1969, pp 382-387. (14) H. Friedberg, Anal. Chem., 27, 305 (1955) (15) R. @&her and H.Freiser, Ed., “Analytical Applicatbns of EDTA and Related Compounds”, Pergamon Press, New York, 1972, pp 204-213. (16) K. Burger, “Organic Reagents in Metal Analysis”, Pergamn Press, Oxford, 1973, pp 118-124. (17) T. T. Gorsuch, “The Destruction of Organic Matter”, Pergamon Press, Oxford, 1970, pp 60-67. (18) F. A. Cotton and G. Wilklnson, “Advanced Inorganic Chemistry, A Comprehensive Text”, Interscience, New York, 1972, p 1044. (19) W. K. Wilmarth and A. Haim, “Peroxide Reaction Mechanisms”, J. 0. Edwards, Ed., Interscience, New York, 1962, pp 175-225. (20) D. A. House, Chem. Rev., 62, 185 (1962). (21) J. M. Anderson and J. K. Kochi, J. Am. Chem. SOC.,92, 1651 (1970). (22) G. V. Bakore and S. N. Joshi, 2. Phys. Chem., 228, 250 (1965). (23) Y. K. Gupta and S. Ghosh, J. Inorg. Nucl. Chem., 11, 320 (1959). (24) P. R. Brontschev, A. Alexiev, and 6. Dimltrova, Talanta, 16, 597 (1969). (25) E. Jasinskiene and E. Jankauskiene, Zh. Anal. Khim., 21, 940 (1966). (26) H. Mueiler, M. Otto, D. Suess, and G. Werner, Chem. Anal., 20, 43 (1975). (27) Y. Ogata and T. Akada, Tetrahedron, 5945 (1970). (28) “Standard Methods for the Examination of Water and Wastewater”, American Public Health Association, Washington D.C., 14th ed., 1976, pp 242-249.
RECEIVED for review August 9,1976. Accepted March 28,1977. Acknowledgment is made to the NSF (Grant No. MPS7505447 and CHE-7617711) and the Oregon State Research Council for partial support of this research. Presented in part at the 172nd National Meeting, American Chemical Society, San Francisco, Calif., August 1976.
