Continuous EDTA Titrations at Low Concentrations W. J. BLAEDEL and R.
H. LAESSIG
Department of Chemistry, University o f Wisconsin; Madison, Wis.
b Continuous analytical techniques including automatic blank correction and direct readout are used in potentiometric EDTA titrations a t very low sample concentrations ( 1 0-W). Application is made to the titration of mixtures of metal ions and of natural water samples for total hardness. Root mean squared deviations a t the 1 O-jM level are 2%# but improve with increasing concentration. The continuous methodology and techniques described in the paper permit titrations a t concentrations where conventional colorimetric and electrometric methods of EDTA titration are not applicable. Limitations of the HgEDTA electrode which arise a t low concentrations are defined and circumvented.
C
POTENTIOMETRIC titration techniques in flow systems have been described ( 1 ) and have been applied t o direct and back EDTA titrations with Hg-EDT.1 electiodes (2-4). hutoniatic blank correction, together with direct readout of sample concentration gave a high degree of automation. The resulting reproducibility and precision indicated that titration of very dilute samples would be possible. I n this study, methods are developed TT hich permit EDTA titrations to be performed on a routine basis a t the l O - 5 ~ i ' concentration level. -1nalyses a t this concentration extend the range of EDTA titrations two orders of magnitude below titration procedures depending on visual determination of colorimetric end points and one order of magnitude below conventional electrometric methods. I n the latter, the diffuse nature of the titration curve and the large titration blanks make determination of the actual equivalence point difficult. The low concentration titration techniques developed in the study are checked by titration of synthetic Ca-Mg mixtures and applied to the titration of natural waters for total hardness. OSTINUOUS
EXPERIMENTAL
Titrator. The equipment and procedure are similar to those described earlier for use a t high concentration ( 3 ) . I n the titrator, shown in block form in Figure 1, the metal ion sample, buffer, and metal ion auxiliary solutions 186
ANALYTICAL CHEMISTRY
53706
are each pumped a t constant rates to mix with a buffered EDTA reagent solution pumped a t a variable rate. The mixed stream passes through the electrodes and out to waste. The difference between the electrode voltage and a standard voltage is applied via the servo system to control the EDTA reagent pump speed, so as to keep the mixed stream always a t the end point. At the end point, the EDTA reagent pump speed (measured digitally with the readout system) is proportional to the sought-for sample concentration. The system is calibrated with standard samples; standardized reagents are not required. The EDTA reagent variable speed pump is constructed so that zero flow rate cannot be attained, and the function of the metal ion auxiliary stream is to offset the minimum EDTA reagent flow rate, so that metal ion samples of zero concentration can be titrated. Electrodes. I n previous studies (2, 3) two types of electrodes could be used satisfactorily for samples down to concentrations of 0.01M: a mercurycoated tubular platinum electrode (hITPE), and a flow-through dropping mercury electrode (DME). The M T P E proved satisfactory for titrating samples down to 10-5V1 but the DME did not. Below 0.00131, the slopes of the titration curves obtained with the DATE fell off badly as sample concentration decreased. The end point (steepest slope) also fell increasingly past the equivalence point as concentration decreased. Figure 2 compares titration curves for the M T P E and DZvIE for the titration of 0.002M Mg(I1) with O.0OlL1.;r EDTA. N T P E and hanging mercury drop (HMD) electrodes with freshly formed surfaces gave titration curves with poor slopes, which increased rapidly in steepness as the surface aged. For each surface most of the change occurred
i
METAL ION SAMPLE Z.Olml/MIN
SOLUTION
t
during the first 20 minutes, with essentially no additional change occurring after an hour. ilfter termination of this aging, the M T P E remained stable over weeks of use without recoating. No further work was done to find whether or not the components of the aqueous solution were involved in the aging. The size of the potential break and the absolute value of the electrode potential also changed during the aging period, but stabilized thereafter. During the electrode studies, it was also found that any mercury in the vicinity of the electrode, but unconnected to it, caused high results in the titration of metal ion samples with EDTA. This error also was observed in batch titrations with a Hg-pool electrode (11). These findings point to the superiority of the M T P E over the D M E for titrations a t low concentrations. All of the following work was done therefore with the M T P E . Performance of Titrations. At low sample concentration, selection of the equivalence point is difficult because of the poor slope of the titration curve and because of impurities in the reagents t h a t contribute to the blank. Figure 3 shows the dependence of the slope of the titration curve on sample concentration. The titrator, by maintaining constant ionic strength, pH, and buffer concentration, and by controlling the concentration of Hg(I1) in the main stream, gives titration curves whose shapes are independent of reagent pump rate and only slightly dependent on sample concentration, This reproducibility allows titration to a pre-selected end point potential. The compensator (3)permits uncritical selection of the end point by automatically correcting for any systematic errors due to a discrepancy between the end point and the correct equivalence point.. By use of standard samples for calibration,
;SOLUTION PATH ELECTRICAL LINK MECHANICAL LINK MIXER
WASTE
METAL ION SOLUTION EDTA REAGENT
Figure 1 .
READOUT SYST
Block diagram of continuous titrator
I
I
I
l
l
a
90 110 PUMP SPEED 9, OF EQUIVALENCE
Figure 2.
Titration curves with MTPE and DME
0.002M Mg(N03)2 titrated with 0.001M EDTA
I I
I
I
I
I
I
I
I
loo 110 % OF EQUIVALENCE
90
errors arising from the conventional reagent impurities (blank) are also compensated for. To prepare the titrator for operation, an EDTA reagent concentration is chosen that is of the same order as the sample concentration. K'ext, the auxiliary metal ion concentration is chosen so t h a t when a blank sample is pumped, the auxiliary metal ion introduced is approximately equivalent to the EDTA reagent a t its minimum pump speed. If, a t the first attempt, the EDTA reagent pump speed is not the lowest 5% of its range, a small dilution or concentration of the auxiliary metal ion solution may be made. At concentrations above 10-4V, this proper auxiliary metal ion concentration and its pump rate may be calculated from the EDTA reagent concentration and its minimum pump rate. At concentrations below 10-4M, blank metallic impurities become important, so less than equivalent amounts of auxiliary metal ion are re-
Figure 3. Dependence of titration curve upon sample concentration Titration conditions given in Table I. Approximate sample and EDTA concentrations indicated for each curve
quired to offset the minimum EDTA reagent pump speed. (See Table I). After adjustment of the auxiliary metal ion solution, 2 to 4 standard samples are used to calibrate the titrator, which is set for automatic compensation of the blank and for direct readout (3). The calibration used to obtain direct readout is dependent upon many factors-solution concentrations, pump rate, end point potential, etc. The titrator is therefore calibrated a t the beginning of each running period without striving to reproduce the previous conditions precisely. Standard samples are run occasionally thereafter as
Table 1.
EDTA reagent molarity 5 x 1x 1x 1x 1x
10-2 10-2 10-3
10-40 10-6"
0.26 0.14 0.038 0.0056 0.0056
0.50 0.25 0.05 0.0125 0.0125
2e
Titration Data for Low Concentration Studies
Auxi1iary
metal ion
Molarity Sample
COPPER(II) SAMPLES 3.7 x 10-2 0.25 to 15 x 7 . 5 x 10-3 0 . 5 to 3 . 5 x 8.1 x 10-4 0.5 to 3.5 x 8 . 0 x 10-5 0 . 5 to 3.5 x 3.0 x 0.5 to 3 . 5 X
10-2
10-2 10-3 10-4 10-6
CALCIUM(II) SAMPLES 0.26 0.50 3.7 x 10-2 0.25 to 15 X 10-a 0.14 0.25 7 . 5 x 10-3 0.5 to 3 . 5 X 0.038 0.05 7.5 x 10-4 0.5 t o 3.5 x 10-3 0.0062 0.01 4.3 x 10-6 0.5 to 3.5 x 10-4 0.0062 0.01 3.5 x 10-6 0 . 5 to 3 . 5 X 10-6 EDTA reagent contains lO-7M Hg(I1). All other EDTA solutions contain lO-@MHg(I1).
x 10-2 1 x 10-2 1 x 10-3 1 x 10-40 1 x 10-6a 5
a
Buffer molarity in EDTA Buffer reagent solution
checks. Usually, even for running periods in excess of 5 hours, there is little loss of calibration. However, when a standard sample does shorn a small loss of calibration, it may be corrected for by a corresponding adjustment of the automatic blank compensator. Reagents. The following stock solutions were prepared from reagent chemicals: 0.05V E D T A (from disodium salt), 1. O X SH3-1 . O M XH4N03 buffer, and 0.05M solutions of various metal salts iCu (K03)2.6H20, Ca (YO3) 4 H ~ 0 and , hIg(hT0~)2.H~O].The metal salt solutions were standardized by potentiometric titration against stand-
No. of samples titrated
R.m.s. error, moles/liter
14 16 12 16 8
0.7 x 10-3 0.6 x 10-4 0.6 X 10-6 1 . 2 x 10-6 2.2 x 10-7
11
0.7 0.7
12 15 13 14
x 10-3 x 10-4 1.1 x 10-5 1 . 5 X 10-8 2 . 1 x 10-7
VOL. 38, NO. 2, FEBRUARY 1966
187
Table II.
Analysis of Ca(ll)-Mg(II) Mixtures
Buffer molarity in Molarity EDTA reagent EDTA Buffer Auxiliary molarity reagent solution metal ion Sample 0.5 to 3.5 X 0.14 0.25 7.5 x 10-3 1 x 10-2 0 . 5 to 3.5 x 10-3 0.038 0.05 8 . 1 x 10-4 1 x 10-3 1 x 10-4a 0.0056 0.0125 8 . 0 x 10-5 0.5 to 3.5 x 10-4 EDTA reagent contains lO''fi!f Hg(I1). All other EDTA solutions contain 10-6M Hg(I1).
No. of samples titrated 29 19 17
R.m.s. error, moles/liter 2.4 x 10-4 1 . 9 x 10-5 2 . 4 X 10-6
5
Analysis of Water Samples
Table 111.
Sample source
Date taken
Madison tap water Madison tap water Madison tap water Lake llendota Lake 11ononz Lake Wingra Steam plantc Softened waterd Private well, Madison
Total hardness, meq./ml. Mercury Continuous pool titration6 titrator" 0.00390 0.00343 0.00370 0.00168 0.00174 0.00164 0.00168 0.00047 0.00313
0.00391 0.00342 0.00370 0.00169 0.00174 0.00164 6,00168 0.00050 0.00313
Average of duplicates. Average of triplicates. c Filtered and chlorinated Lake ,hlendota water supplied to campus buildings. d Madison tap water, from a private home softener. Q
b
ard 0.01043V EDTA solution. With the concentration ranges used in the study, inetallochromic indicators Rere unsatisfactory. The stock solutions 1 1ere diluted volumetrically tvith distilled deionized water. RESULTS AND CONCLUSIONS
Low Concentration Studies. A large number of Ca(I1) and Cu(I1) samples were titrated in the concentration range 0.1 to l O - ~ X . Results are shown in Table I. The root mean square differences given in the table are calculated from the differences between the determined value and the known value for each sample, and are regarded as a measure of the precision of the titrator a t the designated concentration levels. Relative errors, referred to midrange sample concentrations, range from O.5y0a t high concentrations to 2% at low ( i O - 5 M ) sample concentrations. Even though the magnitude of the sampling errors for real samples a t l O - 5 M would probably render the 2y0 error in the titration meaningless, it should be emphasized that the 2y0 error in the titration itself is valid and has meaning for the synthetic samples of Table I. This precision and reproducibility is
188
ANALYTICAL CHEMISTRY
achievable with the continuous titration technique because of the high degree of automation, and because of the isolation of the system from the surroundings, Titration of Mixtures. The DOtentiometric E D T A batch titration of the total metal ion in a sample has been reported previously (5, 8). As shown by Reilley and Schmid (IO) and later discussed in relation to the continuous titrator (2, S), the shape of the titration curve in the region past the equivalence point is independent of the metal ion being titrated. Selection of an end point in this region, but still on the steep portion of the curve, results in a slight over-titration, but this error can be eliminated by calibration. The continuous titrator was applied to a series of Ca(I1)-Mg(I1) mixtures ranging from pure Ca(I1) to pure RSg(II), with the total titratables ranging from 0.01 to 0.0001;CI. All calibrations were performed with standard Ca(I1) solutions. Results are given in Table 11. The root mean square errors are computed from the differences between the determined values and the correct values known from dilution data. The relative root mean squared error is around
2y0for most samples, being only slightly larger than the error for comparable samples containing only a single metal ion. A study of the original data (not shown in Table 11) reveals no bias, indicating that standard mixtures need not be used for calibration even though mixtures are titrated. Titration of Naturai Waters. E D T A titrations of total hardness have been reported previously (5, 8), and are used in commercial titrators (7, 9) and in the brewing industry (6). T o demonstrate the utility of the titrator for water samples, eight different water samples from various sources were titrated using the procedure described in the preceeding section. The results are reported in Table 111. Samples which fell outside the titratable rangei.e. the tap water samples-were volumetrically diluted 2-fold with distilleddeionized water before titration. Total hardness determined with the continuous titrator compares very well with the total hardness determined by batch titration with a mercury pool electrode. LITERATURE CITED
(1) Blaedel, W. J., Laessig, R. H., ANAL. CHEM.36. 1617 11964). (2) Ibid., 37, 332 ('1965): (3) Ibid., p. 1255. (4)Ibid., p. 1650. (5) Fritz, J. S., Garralda, B. B., Ibid., 36, 737 (1964). (6) Goldberg, R. D., Am. SOC.Brewing Chemists Proc. 1962, p. 140. (7) Hach Che,pical Co., "Water Analysis Procedures, Catalog 8, Ames, Iowa. (8) Schoenbaum, R. C., Breckland, E., Talanta 11, 659 (1964). (9) Technicon Controls, Inc., Bulletin H2 (Hardness IIIG) Technicon Gorp. Chauncy, N. Y. (10) Reilley, C. N., Schmid, R. W., ANAL.CHEM.30, 957 (1958). (11) Reilley, C. N., Schmid, R. W., Lamson, D. W., Ibid., p. 953. RECEIVEDfor review October 20, 1965. Accepted November 15, 1965. This work was supported in part by Grant AT( 11-1)1082 from the United States Atomic Energy Commission and in part by a National Science Foundation Summer Fellowship (R.H.L. 1965).
