Null-Point Potentiometric Determination of Fluoride - Analytical

May 1, 2002 - T. A. O'Donnell, and D. F. Stewart. Anal. Chem. , 1961, 33 (3), pp 337–341 ... Donald T. Downing. Analytical Chemistry 1967 39 (2), 21...
0 downloads 0 Views 592KB Size
Null-Point Potentiometric Determination of Fluoride T. A. O'DONNELL and D.

F. STEWART

Deparfment of Chemistry, University o f Melbourne, Parkville, N. 2.) Victoria, Australia

b M a n y existing volumetric methods for fluoride are unsatisfactory. The end point of the alizarin-thorium nitrate titration is indefinite and other methods are inaccurate or not specific for fluoride. A simple and accurate null-point potentiometric method has been developed based on complexing of cerium(lV) b y fluoride and resultant lowering of the cerium(lV)-cerium(ll1) redox potential. Unknown fluoride is added to one cerium(lV)-cerium(ll1) half cell and standard fluoride solution added to a similar half cell until the potential difference is zero. The method has been used to determine fluoride in many binary and complex fluorides and can b e applied to the unbuffered solution from a WillardWinter distillation, to ion exchange eluates, and to pyrohydrolysis distillates. The common anions, chloride, nitrate, and sulfate, do not interfere with the accuracy and specificity of the method.

T

earlier literature on t h e gravimetric and volumetric macrodetermination of fluoride has been surveyed by Busch, Carter, and McKenna ( 1 ) and in great detail by Elving, Horton, and Willard ( 2 ) . The most common of the volumetric methods, titration against thorium or zirconium nitrate (IO), has several disadvantages, a major one being the necessity for careful buffering. The end point is not sharp, particularly when the method is applied after a Willard-Winter distillation, and accuracy is often poor as a result. Acidimetric determinations have been reported recently but usually can only be applied in special cases-e.g., IT-here the fluoride has been recovered from the sample as hydrofluoric acid after pyrohydrolysis (9) or after a n ion exchange process (6,7 ) . I n each of these separation procedures, anions other than fluoride may be converted to the appropriate free acids and so the acidimetric determination which follows the separation step is not necessarily specific for fluoride. Chloride has been shown to introduce gross errors in fluoride determinations after pyrohydrolysis ( 6 ) . The electrometric methods of fluoride determination previously reported have all been inaccurate. An over-all accuracy of 1.8 yo was claimed for a recent conductometric method (3). SevHE

eral attempts to develop a direct potentiometric titration ( 2 ) were based on change of the potential of a redox couple resulting from complexing by fluoride ions of one of the oxidation states of the couple. Despite efforts t o control 501~tion conditions, such as ionic strengths, by saturating the solutions with sodium chloride or adding alcohol, the methods 11ere inaccurate. The method described in this paper, however, gives a n easily determined and reproducible end point and a n accuracy of 0.2% or better. It can be applied with high specificity to a wide range of fluorine-containing samples. As n i t h earlier methods, i t is based on complexing by fluoride of one of the oxidation states of a redox couple but it differs from earlier methods in that the potential difference measured is that betn-een two half cells, each initially containing the same volume of a solution containing cerium(1V) and cerium(II1) ions or iron(II1) and iron(I1) ions. The unknown fluoride solution is added to one half cell and an equal volume of w.ter to the second. The magnitude of the potential difference gives some measure of the fluoride concentration but the accuracy of the method results from adding standard fluoride solution to the second cell until the potential difference is zero. During the addition of the titrant, water is added a t the same rate to the first half cell, so t h a t acidity and ionic strength? remain the same in each half cell. From the initiation of this work, this new analytical method has been referred to in research reports (8) as a null-point potentiometric determination. While the work was in progress, a paper was published (4) which described the microdetermination of chloride by a method called precision null-point potentiometry. The comnionly occurring anions, chloride, bromide, nitrate, and sulfate, do not interfere with the null-point potentiometric determination of fluoride, although acetate does t o a small extent and there is gross interference from oxalate, phosphate, and molybdate. Alkali metals do not interfere and so this method may be applied t o solutions of alkali fluorides, or to neutralized hydrofluoric acid. Where a cation such as uranium affects the accuracy of the determination, the method can be satisfactorily modified t o give accurate results by adding to the second half

cell, before running in the titrant, the same amount of the cation as is contained in the amount of sample added to the first half cell. Frequently the cation can be removed prior to determination of fluoride by a simple chemical step. This procedure will be illustrated later in this paper in the determination of the fluorides of several metals. Alternatively, t o eliminate cation interference, the sample for analysis may be dissolved and passed through a n ion exchange column. Otherwise, the sample may be subjected to pyrohydrolysis or to a Willard-Winter distillation. Fluoride in the eluate or the distillate can then be determined by null-point POtentiometry, which is specific for fluoride so t h a t chloride, sulfate, and other anions in the eluate or distillate do not interfere. The solutions most commonly used in developing the method were 0.05 to 0.1N in fluoride, but using only the simple apparatus required in this concentration range the method still gives accurate results down to 5 x 10-3M. At present, procedures are being modified to use null-point potentiometry for the microdetermination of fluoride. To show that the null-point method is not restricted to fluoride analyses, zt little work was done on analysis of phosphate and oxalate using a n iron(II1)-iron(I1) couple. I n each case, results were promising. EXPERIMENTAL

Apparatus. The method uses as half cells two 250-ml. beakers, connected by a n agar-potassium chloride salt bridge, each beaker being provided with a stirring rod. Platinum wire electrodes, which needed no pretreatment other than conventional cleaning, were connected t o a n inexpensive potentiometer of relatively low sensitivity, in this case a Pye portable potentiometer (Catalog SO. 7569P). It, was felt that inaccuracies might result from the use of acidified fluoride solutions in glass apparatus. Accordingly, a series of determinations was carried out in which the glass of the electrodes, half cells, and salt bridge )vas replaced by polyethylene. However, this precaution proved to be unnecessary. All reagents were of analytical reagent grade. Procedure. For analysis of a solution which is approximately 0.05M in fluoride, prepare a solution by dissolving 12.7 grams of ceric ammonium VOL 33, NO. 3, MARCH 1961

337

sulfate in 200 ml. of n a t e r and 14 ml. of 18M sulfuric acid. Then dissolve 2.8 grams of cerous sulfate and dilute this solution t o 1 liter. The resulting solution is 0.005M in Cez(S04)3-that is, 0.01M in cerium(II1) ions-and i t is 0.01M in cerium(1V) and 0.25M in sulfuric acid. These concentrations are chosen for analysis of a solution approximately 0.05M in fluoride, so that, when the experimental procedures listed below are followed, the ratio of fluoride concentration t o concentrations of cerium (IV) and cerium(II1) will have the optimum value of about 2: 1:1. The concentrations may need modification if the fluoride concentration in the solution for analysis is very different from 0.05M. The relation between fluoride and cerium concentrations nil1 be discussed later. To each half cell, add by pipet 50 ml. of the cerium(1V)-cerium(II1) solution prepared as above. Connect the half cells and check that the e.m.f. of the cell is zero. Pipet a suitable aliquot, about 20 ml., of the unknown fluoride solution into one half cell and add the same volume of distilled water to the second. The magnitude of the e.m.f. at this stage gives an approximate measure of the fluoride concentration in the unknown solution. Add standard sodium fluoride solution (in this case 0.0.5.W) from a buret t o the second half cell. During the titration add distilled water from another buret to the first half cell a t the same rate as the titrant is added. After each addition, stir the solutions thoroughly and measure the potential difference between electrodes. For most accurate determination of the end point-that is, \There the potential difference is zero-measure the e.m.f. of the cell as the titrant is added in steps of about 0.2 ml. for about 1 ml. on either side of the end point. Plot values of e.m.f. against volume of titrant and read from the graph the volume required to produce zero e.m.f. EXPERIMENTAL CONDITIONS

Choice of Redox Couple. Initially, the suitability of the iron(II1)-iron(I1) system as the redox couple in t h e potentiometric determination was investigated and later, the tin(1V)titaniuni(1V)-titanium(III), tin(II), and cerium (IV)-cerium (111) systems were studied. Of these, only the iron and cerium systems gave satisfactory results under simple experimental conditions. The cerium couple has been used for most viork because of the greater stability of the cerium solutions and because the e.m.f. resulting from fluoride addition to one half cell is about five times greater than for the iron system. Hom-ever, i t will be shown later that, for certain special applications of the potentiometric method, the iron couple should be used. Choice of Solution Concentrations. The e.m.f. due to addition of fluoride t o one half cell is almost directly proportional t o the ratio of the con338

ANALYTICAL CHEMISTRY

I

I

I

I

I

1

I

M O L E S F L U O R I D E ION P E R MOLE CERIUM(IV) ION

Figure 1. E m f . for addition of 0 . 0 5 M NaF to solutions initially 0.01M in Ce(lV) and Ce(lll) and 0 . 2 5 M in HzS04

I

w

I

\ I

I

- log

I

[c e3 +I [cc*+]

Figure 2. Effect of change of Ce(lll) concentration on e.m.f. for addition of 2 0 . 0 0 ml. of 0 . 0 5 M NaF to solutions 0.01M in Ce(lV) and 0 . 2 5 M in HzS04

centrations of fluoride and cerium(1V) until a value of 4 for t h e ratio is reached, as shown in Figure 1. For most satisfactory titrations, experimental conditions have been selected so that the ratio of fluoride to cerium(1V) concentrations is in the region of 2 to 3. Figure 2 shows that the relative concentrations of cerium(II1) and cerium (IV) are not very critical. Equal concentrations of the two species are used because under these conditions the measured e.m.f. is greatest; also, a t high cerous concentrations there is danger of precipitation of cerous fluoride while a t low concentrations the resistance of the cell is high and measurement of the e.m.f. is inaccurate with the simple equipment used. If the acidity of the solutions used is not sufficiently high, precipitation of basic ceric compounds will occur, but Figure 3 shows that the sensitivity of the method decreases with increasing acidity. Since dilution of the original solutions in the half cells occurs on addition of neutral fluoride solutions, the

optimum acidity of the cerium(1V)cerium(II1) solution has been fixed a t 0.25-V Iyith respect to sulfuric acid. Figure 3 shows that reliable results may be obtained, with some loss in sensitivity, if a particular system for analysis requires the use of acidities greater than 0,25Min sulfuric acid. RESULTS AND ACCURACY

I n Table I are given three series of determinations. For series I, 20.00 ml. of 0.05-11 sodium fluoride solutioni.e., 19.00 mg. of fluoride-was used as the sample for analysis in each case. The 50 ml. of solution added initially to each half cell was 0.01M in each of cerium(1V) and cerium(II1) and 0.25-If in sulfuric acid. These concentrations ensure the optimum working conditions m-ith a fluoride:cerium(IV) :cerium(III) ratio of 2:l:l. Glass beakers and salt bridges were used. The same experimental conditions were used in series 11, but the amount of sodium fluoride solution taken as the sample for analysis was

Table I. Results of Determinations of Fluoride Added as Sodium Fluoride Fluoride, Mg. Series

No. I

1

I

I

I 2 3 M O L A R I T Y W I T H RESFECT TO

I 4

I

RANGE OF METHOD

The accuracy quoted in the section above can be obtained in analyzing solutions 0.05 to 0.5.11 in fluoride. Fluoride a t lower concentrations can be determined with slightly reduced accuracy by scaling down the concentrations of all reagents so that the optimum fluoride: cerium(1V) : cerium(II1) ratio of 2 : l : l is maintained approximately in 0.25M sulfuric acid solution. I n analyses of solutions 5 X 10-3-lf in fluoride, the accuracy was 0.5 and 1% a t 1 X 10-3Jf, which was the effective lower limit of concentration for the method in the form in which it has been investigated. Choice of the optimum fluoride to cerium concentration ratios presupposes a knowledge of the approximate fluoride concentration in any solution for analysis. The present method can be used crudely to determine the approximate fluoride concentration. Solutions 5 x 10-3-V and 10-3M in fluoride were titrated against 0.05M standard sodium fluoride in solutions 0.01M in cerium (IV) and cerium(II1). Although there

18.97, 19 . O s , 18.97,

31.08

31.11

1'3.00; 19.00; 19.00

HZSOs

Figure 3. Effect of acid concentration on e.m.f. change for addition of 20.00 ml. of 0.05M NaF to cerium solutions

varied. For series 111, polyethylene replaced the glass in contact n-ith acidified fluoride solutions, 19.00 mg. of fluoride being used as sample in each case. For series I and 11, the mean percentage deviations are 0.16 and O.ZO%, respectively, and 0.07y0 for series 111. Generally, there is no great advantage in the use of polyethylene apparatus. I t is possible that, when using all-glass apparatus, errors due to glass attack compensate. To eliminate this possibility, some determinations were made with one glass and one polyethylene half cell. -4lthough the acidified fluoride solution stood in the half cells for some hours before the titration, no errors were introduced.

Found

-

Added 19.00

was no scaling d o m of the cerium concentrations, the analysis mas performed with a n accuracy of 2% in the first case and 10% in the second. Such a rough determination allows selection of the appropriate concentrations of cerium solutions and titrant to ensure accurate analyses. INTERFERING SUBSTANCES

The effects of several different anions on the accuracy of the potentiometric method have been studied. I n each case, the solutions were approximately equimolar in both fluoride and the anion under investigation. Sulfate and nitrate caused no significant interference. Equimolar chloride led to results that were less than 0.15YGtoo high. Chloride interference increased m-ith increasing concentrations. As complexing or chemical reaction with cerium solutions became more significant, anions showed greater interference. Equimolar acetate gave results about 2yc too high, but the strongly complexing osalate, phosphate, and molybdate caused gross interference. I n the case of molybdate, increase in the acidity of the solution-Le., decrease in the concentration of molybdate ions as such-reduced the interference which became less than 1% in a solution 3 M in sulfuric acid. Bromide interfered through reduction of cerium (IV) to cerium(III), but analysis for fluoride in the presence of bromide can be performed using iron(II1)-iron(I1) as the redox system. Of the cations, the alkali metals and ammonium did not interfere. Certain others, such as nickel and cobalt, appeared to have no effect on the determination. In general, it is relatively easy to remove the cation by precipitation or some other process, as will be discussed in the nest section. Some

other cations, such as uranium, interfered to a greater or lesser extent. Frequently the effects of interference from such sources can be removed without separation of the interfering ion by a process of compensation-Le., by adding an appropriate amount of the interfering species to the half cell to which titrant is added. In some instances, the addition must be made accurately as in the case of uranium interference, while in others only an approximate cumpensation need be made. ;In example of the latter case is that in n hich gross amounts of chloride are preqent in a Sample. Approximate compensation suffices because chloride in an amount equimolar with fluoride has no significant effect on the accurac? of the determinntion. APPLICATION TO PARTICULAR SYSTEMS

Cobaltic and Mercuric Fluorides. Fluoride has been determined in each of these fluorinating agents by adding sodium hydroside solution to the weighed sample in a closed vessel and filtering. Aliquots of the filtrate, previously made up to a definite volume, were titrated potentiometrically. Obviously. this procedure can be applied to any sample which yields an insoluble oside or hydroxide. Antimony Trifluoride. Determination of fluoride in a sample of antimony trifluoride, purified by vacuum sublimation, gave the theoretical result. The solid was dissolved in a minimum amount of hydrochloric acid. Antimony n.as precipitated with hydrogen sulfide. Hydrogen peroxide was then used to osidize sulfide in the neutralized filtrate to sulfate which does not interfere with the potentiometric determination. This procedure also has fairly wide application. It would be preferred to that suggested in the paragraph immediately above in cases where addition of sodium hydroxide leads to the formation of oxyVOL. 33, NO. 3, MARCH 1961

* 339

fluorides or of gelatinous hydroxides on which coprecipitation of fluoride might be extensive. Silver Fluorides. Fluoride in simple and complex silver fluorides can be determined directly after precipitation of the silver as chloride. Potassium Hexafluoronickelate (IV). The ratio of fluoride t o nickel in this compound Tvas established by treating a weighed sample n-ith sodium hydroxide solution in a closed glass vessel, boiling the resulting suspension, and acidifying with dilute sulfuric acid. The resulting clear green solution was made up to a definite volume and fluoride was determined on a n aliquot n ithout removal of nickel. Uranium Fluorides. T h e presence of uranium ions leads to fluoride analysis figures which are 5 to 10% too lo^. Interference by uranium can be eliminated by adding to the second half cell before addition of titrant the exact amount of uranium added to the first half cell. Fluoride may then be determined with the usual accuracy. Dental Fluxes (Borate-Fluoride). Fluoride was determined to a n accuracy of about 0.2% in tetraboratefluoride mixtures a n d in commercial dental fluxes. The solids were dissolved in water, the solution was neutralized to phenolphthalein, and aliquots were titrated. Willard-Winter Distillates. Insoluble samples or those which contain species Ivhich m:ll interfere in t h e subsequent titration are frequently distilled n-ith sulfuric or perchloric acid in a glass still and aliquots of the distillate of fluorosilicic acid are titrated against thorium or zirconium nitrates using alizarin compounds as indicators. Control of pH is vital and the aliquot must be buffered very carefully. The end point, a subjective comparison between the intensity of red color in the solution and in a prepared color match, is not easy to detect. The titration reaction is not stoichiometric and a calibration curve must be prepared to calculate fluoride in terms of volume of titrant. The null-point potentiometric method can be used to determine fluoride in a Willard-Tinter distillate. Again, the reaction is not stoichiometric, but a linear relation between fluoride in the distillate and volume of titrant can be obtained with correct choice of experimental conditions. The calibration procedure eliminates the need for neutralization of the distillate and the end point is definite and reproducible. Since sulfate does not interfere, the use of perchloric acid may be avoided in the distillation. Aliquots of neutralized fluorosilicic acid from a Willard-Winter distillation were titrated against sodium 340

0

ANALYTICAL CHEMISTRY

A

/ /

1

I

2 0

IO

/

C/

I

30

ML. SODIUM FLUOROSILICATE Figure 4. Effect of acid concentration on calibration curve for titration of distillate with NaF solution For each solution, Ce(lV) = Ce(ll1) = 0.01 M

A. B. C.

0.25M H2SOd 0 . 5 0 M H&O4 2.OM H.S04

ML, SODIUM FLUOROSILICATE Figure 5. Effect of variation in Ce(lV) concentration on calibration curve for titration of distillate with 0.05M NaF For each solution, Ce(ll1) = 0.01 M; HiSOa = 0.25M A. 0.04M Ce(lV1 8. 0.01M Ce(lV1

fluoride using the null-point method. The departure from direct proportionality between the amount of fluorosilicnte and volume of titrant became greater ~ i t hcontinued increase in the fluorosilicate concentration-i.e., as the amount of cerium(1V) in ewess decreased. a l . ~ , the departure from a stoichiometric reaction became greater as the acitlity of the solution as increased. This latter effect is shown in Figure 4, n-hile Figure 5 shons the effect of increaaing the ratio of ceriuni(1V) to ceriumiII1). W i e n the solutions :ititled to the half cells are 0.02.11 in crrium(1V). 0.01V in ceriuni(III), and 0.25.11 in wlfuric acid, the calibration curve is a straight line for addition of 0 to 30 ml. of 0.0LlI fluoride, xvhen the strength of the fluorosilicate solution is comparable n i t h that of the titrant. An accuracy of about 0.2% can be obtained in a fluoride determination after a TTillnrd-Winter distillation by :idopting thc procedure in which the

usual ratio of the concentration of cerium(1V) to cerium(II1) is doubled. Solutions from Pyrohydrolysis and I o n Exchange Separation. These separation processes give n distillate or a n eluate which, in t h e ideal case, is a solution of hydrofluoric acid. The normal procedure is to determine the fluoride acidimctrically. Because of the weakness of hydrofluoric acid, the end point can lack sharpness; but the real difficulty in these procedures is that the presence in the sample of anions other than fluoride can lead to high results-e.g., if chloride is present, unknown amounts of hydrochloric acid may be produced. The null-point potentiometric method may be used with great advantage t o determine fluoride in a neutralizrd aliquot of the distillate or eluate, because the method is specific for fluoride and. unlike other specific volumetric niethotls for fluoride, provides a definite and reproducible end point.

LITERATURE CITED

( I ) Busch, G. W., Carter, R. C., McKenna, F. E., in “Analytical Chemistry

of Manhattan Project,” C. J. Rodden, ed., Vol. 1, Chap. 5, bIcGraw-Hill, New York, 1950. (2) Elving, P. J., Horton, C. A., Willard, H. H., in “Fluorine Chemistry,” J. H. Simons, ed., Vol. 11, Chap. 3, Academic Press, New York, 1954. (3) Kubota, H., Surak, J. G., ANAL. CHEW31, 283 (1959). (4) Malmstadt, H. V., \%‘inefordner,J. D , Anal. Chim. Acta 20, 283 (1959). (5) ShEhyn, H., ANAL. CHEM.29, 1466 ( 195i). (6) Silverman, H. P., Bowen, F. J., Ibid., 31, 1960 (1959). (7) Sporek, X. F., Ibzd., 30,1030 (1958). (8) b niversity of Melbourne, Victoria, Australia, Rept. of Research and Investigation (1958 and 1959). (9) Warf, J. C., Cline, W. D., Tevebaugh, R. D., ASAL. CHEY.26,342 (1954). (10) Willard, H. H., Winter, 0. D., IND. ESG.CHEM., h A L . ED. 5 , 7 (1933). RECEIVEDfor review June 6, 1960. Accepted December 7 , 1960.

Potentiometric Determination of Chlorine Bound to Boron in Mixtures Containing ZChlorovinyl Boron Chlorides and Ethyl Boron Chlorides H. G. NADEAU, D. M. OAKS, and R. D. BUXTON Olin Mathieson Chemical Corp., New Hoven, Conn. b An accurate method i s described for the determination of chlorine bound to boron in organoboron compounds containing boron-chlorine and carbon-chlorine bonding. The method has been tested and used for compounds and mixtures of dichloro-2chlorovinylborane, chlorobis-2-chlorovinylborane, tris-2-chlorovinylboraneI dichloroethylborane, chlorodiethylborane, chloro-2-chlorovinylethylborane, and boron trichloride. The method i s highly reproducible and shows good accuracy over a wide range of concentration.

A

for chlorine bound to boron in mixtures containing compounds which possess both chlorine bound to lioron and to carbon necessitates a method which distinguishes each type of chlorine. While procedures are available for the determination of chlorine in organoboron compounds (Parr bomb-sodium peroxide fusion, T’olhard and Carius methods), the values arrived a t give total chlorine ULPSIS

content. Aqueous hydrolysis of chloro2-chlorovinylboranes, with subsequent titration of hydrochloric acid with either base or silver nitrate, fails to differentiate between chlorine bound to boron and to carbon, because these compounds break down in aqueous medium (above p H 3) to form hydrochloric acid, orthoboric acid, and acetylene. The hydrolysis of chlorine bound to carbon proceeds at a much slower rate than that for the chlorine bound to boron, and various attempts have been made to stabilize the 2chlorovinylboric acid formed. If the compound was dissolved in benzene and titrated with a solution of potassium methylate in benzene containing a small amount of methanol, only chlorine bound to boron was determined. The titration was carried out to a thymol blue end point. It was necessary, however, for the analysis to be rapid since on standing some breakdown of the 2-chlorovinylboric acid occurred. This procedure was found effective for chlorovinylboranes, but if sample mixtures contained chloro-

ethylboranes, the method was inaccurate. Under these conditions all of the chlorine bound to boron in the chloroethylboranes did not dissociate. Quantitative aqueous hydrolyses of these compounds are well known (a), and have been performed by the authors many times. Apparently, the poor recovery of chlorine obtained when dealing with mixtures containing alkyl boron chlorides by the benzene potassium methylate titration is related to the low polarity of the solvent. In the method presented, a polar solvent system has been chosen, and a means of stabilizing the chlorine-carbon bonds has been found. The compounds, or mixture of compounds containing both types of chlorine, are dissolved in a strong nitric acid-methanol solution. Hydrochloric acid formed through esterification of the chlorine bound to boron is titrated potentiometrically with standard silver nitrate. Chlorine bound to boron in the chloro-2-chlorovinylboranes and in the chloroethylboranes esterifies quantitatively; furthermore, the 2VOL. 33, NO. 3, MARCH 1961

341