Polarographic Microdetermination of Fluoride A. M. Bond and T. A. O’Donnell Department of Inorganic Chemistry, University of Melbourne, Victoria 3052, Australia. A rapid, highly-specific polarographic method, based on s hift of the u ra niu m(V)-u r a niu m (I I I) haIf -w a ve potential, has been developed for precise determination of fluoride in the concentration range 2.5 X to 1O-W (0.05 to 20 ppm). Because the method requires no prior separation or concentration of the fluoride, it is particularly applicable to analysis of potable waters. With simple modification, it can be used to determine higher fluoride concentrations and it has been applied successfully to the analysis of a wide range of inorganic fluorides.
To DATE MOST microdeterminations of fluoride have been based on spectrophotometric procedures, such as the zirconiumalizarin or the cerium(II1)-complexan method. I n general these methods are empirical and extremely sensitive to small changes in experimental conditions ( I ) . They are also subject to greater or less interference from species other than fluoride in the sample and frequently prior separation of the fluoride by a process such as micro-diffusion or distillation is necessary ( 2 ) . Recently a lanthanum fluoride membrane electrode has been reported (3) as being specific in its reaction to fluoride and sensitive to very small fluoride concentrations. Lingane ( 4 ) has used it to monitor change in fluoride concentration at the equivalence point of a macrodetermination of fluoride based on titration, and Durst and Taylor (5) have applied it to microdetermination of fluoride. O’Donnell and Stewart modified (6) their null-point potentiometric method (7) which was based on complexing by fluoride in the cerium(1V)-cerium(II1) system. However, it was not possible to obtain reliable results at fluoride concentrations below 5 x lO-4M(lO ppm), because of electrode instability at cerium concentrations below lO-3M. Polarography seemed a possible solution to the problem, because low metal ion concentrations can be detected with accuracy. Also, DeFord and Hume (8) have derived an equation for a complex ion in solution relating half-wave potential, diffusion current, and concentration of ligand. Several polarographic systems, such as the reduction of tin(II), lead(II), iron(III), calcium(II), vanadium(V), and vanadium(II1) were investigated for fluoride analysis. Several of these showed observable shifts in half-wave potentials at relatively high fluoride concentrations ; but the effects were negligible for fluoride levels less than 10-4M or were not reproducible. The most favourable system involved reduction of a uranium(V1) solution. Kolthoff and Lingane (9) report two
(1) C . K. Lim, Ana[yst, 87, 197 (1962). (2) C. A. Horton, “Treatise on Analytical Chemistry,” Part 11, Vol. 7, I. M. Kolthoff and P. J. Elving, Eds., Interscience, New York, 1961, pp. 268-279. (3) M. S. Frant and J. W. Ross, Jr., Science, 154, 1553 (1966). (4) J. J. Lingane, ANAL.CHEM.,39, 881 (1967). ( 5 ) R. A. Durst and J. K. Taylor, Zbid.,39, 1483 (1967). (6) T. A. O’Donnell and D. F. Stewart, Zbid.,34, 1347 (1962). (7) T. A. O’Donnell and D. F. Stewart, Zbid.,33, 337 (1961). (8) D. D. DeFord and D. N. Hume, J. Am. Chem. SOC.,73, 5321 (1951). (9) I. M. Kolthoff and J. J. Lingane, “Polarography,” Vol. I1 2nd ed. Interscience, New York, 1952, p. 462.
waves in moderately acid solution of uranium solutions, with El,* values of -0.18 and -0.92 cs. SCE. They state that these values are independent of acidity, uranyl concentration or concentration of the supporting electrolyte, potassium chloride, for all values of the latter up to 0.5M. They state that the first wave corresponds with reduction of uranium(V1) to uranium(V) and the second is a n irreversible two-electron reduction to uranium(II1). A recent report ( I O ) gives a value for -1.25 volt for the U(V)-U(II1) half-wave potential in 1M potassium chloride. We find this to be the value at very low acidity; but as the acid concentration is increased, the value changes towards that given by Kolthoff and Lingane. For our purposes, the absolute value is relatively unimportant, as we are concerned only with changes in the half-wave potential. The present work has shown that the first wave is only moderately sensitive to fluoride but that the uranium(V)uranium(II1) wave shows marked changes in the values of El,* and id for very small fluoride concentrations. Shifts of El/*of up to 80 mV were observed for fluoride at the level of 5 X lO-5M(l ppm). Fluoride can be determined in different solutions at this level with a reproducibility of and can be detected with reasonable precision to 2.5 X 10-6M(0.05 ppm). Chemical interference with polarographic determination of fluoride could arise from anions which complex with uranium(V) or from cations which complex fluoride more strongly than does uranium(V). Potentially, the method is also subject to interference from species with reduction waves close to that for uranium(V). However, even when values of El/zare close, A.C. polarography usually allows separate examination of the uranium(V) reduction. Obviously interfering cations could be removed by a preliminary ion-exchange separation, but the simplest procedure would be to shift the uranium(V) wave by changing the acidity. It was found that, in general, interference from other anions and cations was slight even without modification of the basic method. In instances where it did occur, it could usually be eliminated by a process of “saturating-out,” that is by including the species in the supporting electrolyte at a concentration of about 0.01M during the calibration so that the amount in the sample has no effect on the final analysis of fluoride. Phosphate, which occurs with fluoride in many biological, agricultural and mineralogical systems, presents great interference problems in most methods of analysis for fluoride. In this instance, it was found that, at low acidity, phosphate had a similar effect on the value of E l i 2for U(V)-U(II1) to that for fluoride. However, even the effect of a 150-fold excess of phosphate over fluoride could be eliminated by increasing the acidity of the solution. A wide range of metallic fluorides has been analyzed with the polarographic method and the method has been applied successfully to the determination of fluoride in raw and fluoridated drinking waters. Although it is reported as a micromethod, it can be modified very easily to determine fluoride concentrations up to 10-3M.
+2z
(IO) C . Grether, “Metrohm Application Bulletin,” A36E, Jan 4,
1965. VOL. 40, NO. 10, AUGUST 1968
1405
2.50
-
2.0
-
j 1.50
-
kn
were used to give a linear change in the values of Elizfor the U(VkU(II1) wave for the fluoride concentrations expected in samples for analysis. Uranium(V1) solutions from 10-3 to 10-5M were prepared from uranyl nitrate. Well-formed polarographic waves were observed at these concentrations, maxima and distorted waves occurring at higher concentrations. For maximum sensitivity to low concentrations of fluoride, uranium(V1) concentrations between 5 x 10-5 and 10-5M were used. The supporting electrolyte was usually 0.01 to 0.5M in potassium chloride with added hydrochloric acid. For very low fluoride levels, acidity was fixed at about lO-SM, where shift of El/zwas greatest. Presumably as acidity increases, dissociation of hydrogen fluoride is repressed and the effect of free fluoride ion on the reduction of uranium(V) is decreased. To ensure constant conditions during analyses, a bulk solution containing appropriate concentrations of uranium(VI) and supporting electrolyte was prepared and aliquots were used for the calibrations and the determinations. It was found in practice that solutions were stable and polarographic conditions were reproducible so that frequent calibration was not necessary.
C
0
m
.-0 8
G
1.0
-
DISCUSSION
-
0.50
1.120
1.160 1.200 E t US. Ag/AgCI (volts)
1.240
Figure 1. Typical calibration curve for polarographic microdetermination of fluoride Solution lO-5M in U(V1) and 10-*M in supporting electrolyte
EXPERIMENTAL
Apparatus. Most of the dc polarography was studied with a Metrohm Polarecord E 261. The Metrohm ac Modulator E393 was used in conjunction with the Polarecord for all ac polarographic work. For rapid polarography a Metrohm Polarographie Stand E 354 was used with controlled drop times of between 0.16 and 0.32 seconds and scan rates of potential of 1 volt per minute. Silver/silver chloride reference electrodes were used for both ac and dc polarography. All results reported in this paper were measured using rapid polarographic techniques. Some preliminary dc work was carried out on relatively simple polarographic instruments with calomel reference electrodes and ordinary dropping rates of 3-5 seconds for the DME. The instruments used were a Cambridge General Purpose Polarograph and a Cambridge Pen Recording Polarograph. With these instruments, fluoride analyses were less precise and required longer time than with the Polarecord, but they demonstrated that only standard polarographic equipment is necessary to apply the method. All chemicals used were of reagent grade purity except uranyl nitrate which was purified by recrystallization. No maxima were encountered so that maximum suppressors were not added. Oxygen-free nitrogen was used to de-aerate the solutions. Usually the half-wave potentials were estimated using the simple method of Frank and Hume (11). Procedure. Values of El12for uranium(V) reduction were measured at different fluoride concentrations to provide a calibration curve for analysis of unknown fluoride solutions. A typical calibration curve is shown in Figure 1. Suitable concentrations of uranium(V1) and of supporting electrolytes (11) R. E. Frank and D. N. Hume,J . Am. Chem. SOC.,75, 1736
(1953).
1406
0
ANALYTICAL CHEMISTRY
Measurement of Eli2 or id. It was decided to use change in the values of Eli2rather than of id as a measure of fluoride concentration because the former was easier to measure, more reproducible, and gave a linea- calibration graph at low fluoride concentrations whereas id values showed marked curvature in the calibration graph. The most significant reason, however, was that values for U(V)-U(II1) were found to be independent of capillary characteristics of the DME and of small variations in uranyl concentration. Values of id varied markedly with small variations in uranyl concentration. Use of Eli*values meant that there was no error introduced into fluoride analysis as a result of inaccuracy in preparing uranyl solutions. Concentrations, Acidity, Ionic Strength, and Temperature. Optimum uranium(V1) concentrations for well-formed reproducible waves were in the range to 10-5M. The value of El12varied from -1.3 V at very low acidity to -0.93 V at acidity above 0.1M. The uranium(V1) solution should be kept acidic with hydrochloric to prevent hydrolysis, but maximum sensitivity to fluoride is observed at low acidity. Potassium chloride is by far the most satisfactory supporting electrolyte. Other supporting electrolytes were tested; but definition of the wave and magnitude of the shift in El/2 are best in the presence of chloride ion. The method shows little sensitivity to relatively small changes in temperature and ionic strength. As long as ambient temperatures did not vary by more than =t2' C, it was not necessary to use a thermostat. To overcome ionic strength effects, bulk solutions containing acidified uranium(VI) were prepared. The supporting electrolyte was usually to 10-lM in potassium chloride and to 10-*M in acid. Under these conditions, a shift in Ell2 of up to 80 mV was observed for solutions containing 1 ppm of fluoride. Fluoride could be detected down to about 10-8M. Fluoride concentrations above about 10-4M cannot be determined without modification of the procedure because the calibration curve rises very sharply. If however the acidity of the solution is increased to about O.lM, the sensitivity is greatly reduced and fluoride concentrations up to 10-3M(20 ppm) can be determined accurately. At much higher fluoride levels, the U(V)-U(II1) wave cannot be observed at all. AC Polarography. Some simplifications can be introduced with ac polarography. Shifts in summit potentials (EB) instead of half-wave potentials are observed. If interfering
~
species are present with reduction waves near that for U(V) to U(III), it is easier to observe differences in E, than in EIj2. Also the actual value of E, is easier to fix than for El,*. Another advantage is that removal of oxygen by nitrogen bubbling is not necessary with ac polarography. However it was found that, for a given fluoride concentration, smaller shifts occur for E. values than for El,*, that is a smaller degree of sensitivity and precision is possible. I n practice, ac polarography was best applied in the fluoride concentration range to 10-3M. It was observed that, for U(V)-U(II1) reduction, values of summit potentials from ac polarography did not always correspond with half-wave potentials. This indicates some measure of irreversibility. The high values observed for the half-band width of ac polarograms and of E1/4-E3,4values of dc polarograms, which should be 45 mV and 30 mV, respectively, for reversible two-electron reduction, indicate also that nonreversible reduction is occurring under conditions of rapid polarography. The fact that, contrary to earlier beliefs, a n ac wave can be observed for irreversible reduction has been established, recently, by Smith and McCord (12). Hence observations made in our work are not inconsistent with the statement by Kolthoff and Lingane (9) that the system is irreversible. Form of Calibration Curve. Figure 1 shows a typical plot of El,* values for U(V)-U(II1) at different fluoride concentrations. At low fluoride levels, the plot is linear. With increasing fluoride concentration, curvature becomes quite marked, the graph rising very steeply a t higher fluoride levels. This is the first report of a calibration curve of this type for analytical purposes. However Schaap, Davis, and Nebergall (13) have reported a study of the stability of fluoro-stannate(11) complexes. They measured El/z values for tin(I1) reduction a t varying fluoride concentrations. If their values are plotted, a curve very similar in shape to that in Figure 1 is obtained, although much higher fluoride concentrations are involved in the study. RESULTS AND ACCURACY
Ten different solutions containing 1 ppm of fluoride and ten containing zero fluoride were analyzed with a solution lOVM in uranium (VI) and 5 X 10-*M in a supporting electrolyte of potassium chloride and nitric acid. For the fluoride solutions, values of -El;* of 1.1860, 1.1862, 1.1868, 1.1886, 1.1852, 1.1864, 1.1880, 1.1872, 1.1864,and 1.1882 wereobtained. The mean of these values is 1.1869 V with a standard deviation of 0.001 1 V. For the ten zero fluoride solutions, the mean value was 1.1152 V with a standard deviation of 0.0008 V. The percentage standard deviation for a shift of 71.7 mV in El,*is 2%. I n this section of the work, great trouble was taken to measure Elj2 to 0.1 mV. I n normal analysis, it is necessary to measure to 1 mV only. INTERFERING SUBSTANCES
The possibility of chemical interference with this analytical method by complexing has been discussed in the introduction. (12) D. E. Smith and T. G. McCord, ANAL.CHEM., 40,474 (1968). (13) W. B. Schaap, J. A. Davis, and W. H. Nebergall, J. Am. Chern. SOC.76, 5226 (1954).
Table I. Interfering Species Limiting concentration (mole 1-I)
Species c1-
Na+, K+
10-2 10-3
NOS-, ClodBr-, I-
10-3
Cd2+,NH4+, Hg2-, Mo(V1)
10-4
~ 03 ~ -
10-5
Q
10-4
Extent of interference Nonea Nonea Nonea Nonea Nonea Markedb
At level of accuracy of determination
* At low acidity.
Phosphate interference removed by increasing
acidity
The effects of some typical species encountered at microanalytical level are given in Table I. I t was found that relatively few species did interfere and that, in general, interference could be removed by “saturatingout” the effect. That is, a relatively high concentration (about 10-2M) of the interfering species was added to the supporting electrolyte in the calibration so that the effect of the species in the actual analysis was made negligible. Also the concentration of the supporting electrolyte can be deliberately increased to eliminate ionic strength effects if the sample for analysis contains high concentrations of ionic species other than fluoride. If species are present with values of El/*close to that for U(V)-U(II1) under the given conditions, ac polarography allows easier distinguishing of E. values than for El,*values. Alternatively, the value of Ellzfor U(V)-U(II1) can be shifted deliberately from that of an interfering species because of the marked dependence of the former on acidity. Phosphate Interference. At low acidity, the effect of phosphate on the U(V)-U(II1) wave was found t o be comparable in magnitude with that of fluoride. The saturation method of compensation was inapplicable, because the U(VFU(II1) wave was shifted into the hydrogen wave. However, it was established that accurate fluoride analyses can be performed in the presence of a 150-fold excess of phosphate by using a supporting electrolyte 0.1 to 1M in both hydrochloric acid and potassium chloride. I t is suggested that protonation reduces the complexing effect of phosphate. A similar effect has been observed (14) in potentiometric determination of mixtures of phosphate and fluoride. Increase in acidity causes a loss in sensitivity for fluoride detection. At these high acid levels, l ppm of fluoride causes a shift in El/?of about 5 mV. Hence 1 ppm is now about the lowest level of estimation. Also great increase in acidity can cause some overlap of the hydrogen and the U(V)-U(II1) waves. AC polarography can be used to minimize this complication.
RECEIVED for review October 23, 1967. Accepted February 5, 1968. (14) T. A. O’Donnell and P. Tse, Univ. of Melbourne (1965-66) unpublished work.
VOL. 40, NO. 10, AUGUST 1968
1407